RandomWOW parameters

Fix typo for M1 Mac build

CMakeLists.txt

@@ -153,7 +153,7 @@ if(ARM_ID STREQUAL "aarch64" OR ARM_ID STREQUAL "arm64" OR ARM_ID STREQUAL "armv
     src/jit_compiler_a64.cpp)
   # cheat because cmake and ccache hate each other
   set_property(SOURCE src/jit_compiler_a64_static.S PROPERTY LANGUAGE C)
-  set_property(SOURCE src/jit_compiler_x86_static.S PROPERTY XCODE_EXPLICIT_FILE_TYPE sourcecode.asm)
+  set_property(SOURCE src/jit_compiler_a64_static.S PROPERTY XCODE_EXPLICIT_FILE_TYPE sourcecode.asm)
 
   # not sure if this check is needed
   include(CheckIncludeFile)

src/aes_hash.cpp

@@ -175,10 +175,10 @@ template void fillAes1Rx4<false>(void *state, size_t outputSize, void *buffer);
 //key0, key1, key2, key3 = Blake2b-512("RandomX AesGenerator4R keys 0-3")
 //key4, key5, key6, key7 = Blake2b-512("RandomX AesGenerator4R keys 4-7")
 
-#define AES_GEN_4R_KEY0 0x99e5d23f, 0x2f546d2b, 0xd1833ddb, 0x6421aadd
-#define AES_GEN_4R_KEY1 0xa5dfcde5, 0x06f79d53, 0xb6913f55, 0xb20e3450
-#define AES_GEN_4R_KEY2 0x171c02bf, 0x0aa4679f, 0x515e7baf, 0x5c3ed904
-#define AES_GEN_4R_KEY3 0xd8ded291, 0xcd673785, 0xe78f5d08, 0x85623763
+#define AES_GEN_4R_KEY0 0xcf359e95, 0x141f82b7, 0x7ffbe4a6, 0xf890465d
+#define AES_GEN_4R_KEY1 0x6741ffdc, 0xbd5c5ac3, 0xfee8278a, 0x6a55c450
+#define AES_GEN_4R_KEY2 0x3d324aac, 0xa7279ad2, 0xd524fde4, 0x114c47a4
+#define AES_GEN_4R_KEY3 0x76f6db08, 0x42d3dbd9, 0x99a9aeff, 0x810c3a2a
 #define AES_GEN_4R_KEY4 0x229effb4, 0x3d518b6d, 0xe3d6a7a6, 0xb5826f73
 #define AES_GEN_4R_KEY5 0xb272b7d2, 0xe9024d4e, 0x9c10b3d9, 0xc7566bf3
 #define AES_GEN_4R_KEY6 0xf63befa7, 0x2ba9660a, 0xf765a38b, 0xf273c9e7
@@ -197,10 +197,6 @@ void fillAes4Rx4(void *state, size_t outputSize, void *buffer) {
 	key1 = rx_set_int_vec_i128(AES_GEN_4R_KEY1);
 	key2 = rx_set_int_vec_i128(AES_GEN_4R_KEY2);
 	key3 = rx_set_int_vec_i128(AES_GEN_4R_KEY3);
-	key4 = rx_set_int_vec_i128(AES_GEN_4R_KEY4);
-	key5 = rx_set_int_vec_i128(AES_GEN_4R_KEY5);
-	key6 = rx_set_int_vec_i128(AES_GEN_4R_KEY6);
-	key7 = rx_set_int_vec_i128(AES_GEN_4R_KEY7);
 
 	state0 = rx_load_vec_i128((rx_vec_i128*)state + 0);
 	state1 = rx_load_vec_i128((rx_vec_i128*)state + 1);
@@ -210,23 +206,23 @@ void fillAes4Rx4(void *state, size_t outputSize, void *buffer) {
 	while (outptr < outputEnd) {
 		state0 = aesdec<softAes>(state0, key0);
 		state1 = aesenc<softAes>(state1, key0);
-		state2 = aesdec<softAes>(state2, key4);
-		state3 = aesenc<softAes>(state3, key4);
+		state2 = aesdec<softAes>(state2, key0);
+		state3 = aesenc<softAes>(state3, key0);
 
 		state0 = aesdec<softAes>(state0, key1);
 		state1 = aesenc<softAes>(state1, key1);
-		state2 = aesdec<softAes>(state2, key5);
-		state3 = aesenc<softAes>(state3, key5);
+		state2 = aesdec<softAes>(state2, key1);
+		state3 = aesenc<softAes>(state3, key1);
 
 		state0 = aesdec<softAes>(state0, key2);
 		state1 = aesenc<softAes>(state1, key2);
-		state2 = aesdec<softAes>(state2, key6);
-		state3 = aesenc<softAes>(state3, key6);
+		state2 = aesdec<softAes>(state2, key2);
+		state3 = aesenc<softAes>(state3, key2);
 
 		state0 = aesdec<softAes>(state0, key3);
 		state1 = aesenc<softAes>(state1, key3);
-		state2 = aesdec<softAes>(state2, key7);
-		state3 = aesenc<softAes>(state3, key7);
+		state2 = aesdec<softAes>(state2, key3);
+		state3 = aesenc<softAes>(state3, key3);
 
 		rx_store_vec_i128((rx_vec_i128*)outptr + 0, state0);
 		rx_store_vec_i128((rx_vec_i128*)outptr + 1, state1);

src/asm/configuration.asm

@@ -15,7 +15,7 @@ RANDOMX_SCRATCHPAD_L2 EQU 262144t
 RANDOMX_SCRATCHPAD_L1 EQU 16384t
 RANDOMX_JUMP_BITS EQU 8t
 RANDOMX_JUMP_OFFSET EQU 8t
-RANDOMX_FREQ_IADD_RS EQU 16t
+RANDOMX_FREQ_IADD_RS EQU 25t
 RANDOMX_FREQ_IADD_M EQU 7t
 RANDOMX_FREQ_ISUB_R EQU 16t
 RANDOMX_FREQ_ISUB_M EQU 7t
@@ -29,19 +29,19 @@ RANDOMX_FREQ_IMUL_RCP EQU 8t
 RANDOMX_FREQ_INEG_R EQU 2t
 RANDOMX_FREQ_IXOR_R EQU 15t
 RANDOMX_FREQ_IXOR_M EQU 5t
-RANDOMX_FREQ_IROR_R EQU 8t
-RANDOMX_FREQ_IROL_R EQU 2t
+RANDOMX_FREQ_IROR_R EQU 10t
+RANDOMX_FREQ_IROL_R EQU 0t
 RANDOMX_FREQ_ISWAP_R EQU 4t
-RANDOMX_FREQ_FSWAP_R EQU 4t
-RANDOMX_FREQ_FADD_R EQU 16t
+RANDOMX_FREQ_FSWAP_R EQU 8t
+RANDOMX_FREQ_FADD_R EQU 20t
 RANDOMX_FREQ_FADD_M EQU 5t
-RANDOMX_FREQ_FSUB_R EQU 16t
+RANDOMX_FREQ_FSUB_R EQU 20t
 RANDOMX_FREQ_FSUB_M EQU 5t
 RANDOMX_FREQ_FSCAL_R EQU 6t
-RANDOMX_FREQ_FMUL_R EQU 32t
+RANDOMX_FREQ_FMUL_R EQU 20t
 RANDOMX_FREQ_FDIV_M EQU 4t
 RANDOMX_FREQ_FSQRT_R EQU 6t
-RANDOMX_FREQ_CBRANCH EQU 25t
+RANDOMX_FREQ_CBRANCH EQU 16t
 RANDOMX_FREQ_CFROUND EQU 1t
 RANDOMX_FREQ_ISTORE EQU 16t
 RANDOMX_FREQ_NOP EQU 0t

src/asm/program_prologue_linux.inc

@@ -15,6 +15,7 @@
 	mov rsi, rdx                ;# uint8_t* scratchpad
 
 	mov rax, rbp
+	ror rbp, 32
 
 	;# zero integer registers
 	xor r8, r8

src/asm/program_prologue_win64.inc

@@ -28,6 +28,7 @@
 	mov rbx, r9                 ;# loop counter
 
 	mov rax, rbp
+	ror rbp, 32
 
 	;# zero integer registers
 	xor r8, r8

src/asm/program_read_dataset.inc

@@ -1,17 +1,16 @@
+	mov ecx, ebp                       ;# ecx = ma
+	and ecx, RANDOMX_DATASET_BASE_MASK
+	xor r8, qword ptr [rdi+rcx]
+	ror rbp, 32                        ;# swap "ma" and "mx"
 	xor rbp, rax                       ;# modify "mx"
 	mov edx, ebp                       ;# edx = mx
 	and edx, RANDOMX_DATASET_BASE_MASK
 	prefetchnta byte ptr [rdi+rdx]
-	ror rbp, 32                        ;# swap "ma" and "mx"
-	mov edx, ebp                       ;# edx = ma
-	and edx, RANDOMX_DATASET_BASE_MASK
-	lea rcx, [rdi+rdx]                 ;# dataset cache line
-	xor r8,  qword ptr [rcx+0]
-	xor r9,  qword ptr [rcx+8]
-	xor r10, qword ptr [rcx+16]
-	xor r11, qword ptr [rcx+24]
-	xor r12, qword ptr [rcx+32]
-	xor r13, qword ptr [rcx+40]
-	xor r14, qword ptr [rcx+48]
-	xor r15, qword ptr [rcx+56]
+	xor r9,  qword ptr [rdi+rcx+8]
+	xor r10, qword ptr [rdi+rcx+16]
+	xor r11, qword ptr [rdi+rcx+24]
+	xor r12, qword ptr [rdi+rcx+32]
+	xor r13, qword ptr [rdi+rcx+40]
+	xor r14, qword ptr [rdi+rcx+48]
+	xor r15, qword ptr [rdi+rcx+56]
 	

src/asm/program_read_dataset_sshash_init.inc

@@ -8,10 +8,10 @@
 	mov qword ptr [rsp+16], r13
 	mov qword ptr [rsp+8], r14
 	mov qword ptr [rsp+0], r15
-	xor rbp, rax                       ;# modify "mx"
 	ror rbp, 32                        ;# swap "ma" and "mx"
-	mov ebx, ebp                       ;# ecx = ma
-	and ebx, RANDOMX_DATASET_BASE_MASK
-	shr ebx, 6                         ;# ebx = Dataset block number
+	xor rbp, rax                       ;# modify "mx"
+	mov rbx, rbp                       ;# ebx = ma
+	shr rbx, 38
+	and ebx, RANDOMX_DATASET_BASE_MASK / 64 ;# ebx = Dataset block number
 	;# add ebx, datasetOffset / 64
 	;# call 32768

src/common.hpp

@@ -67,7 +67,7 @@ namespace randomx {
 	constexpr int wtSum = RANDOMX_FREQ_IADD_RS + RANDOMX_FREQ_IADD_M + RANDOMX_FREQ_ISUB_R + \
 		RANDOMX_FREQ_ISUB_M + RANDOMX_FREQ_IMUL_R + RANDOMX_FREQ_IMUL_M + RANDOMX_FREQ_IMULH_R + \
 		RANDOMX_FREQ_IMULH_M + RANDOMX_FREQ_ISMULH_R + RANDOMX_FREQ_ISMULH_M + RANDOMX_FREQ_IMUL_RCP + \
-		RANDOMX_FREQ_INEG_R + RANDOMX_FREQ_IXOR_R + RANDOMX_FREQ_IXOR_M + RANDOMX_FREQ_IROR_R + RANDOMX_FREQ_IROL_R + RANDOMX_FREQ_ISWAP_R + \
+		RANDOMX_FREQ_INEG_R + RANDOMX_FREQ_IXOR_R + RANDOMX_FREQ_IXOR_M + RANDOMX_FREQ_IROR_R + RANDOMX_FREQ_ISWAP_R + \
 		RANDOMX_FREQ_FSWAP_R + RANDOMX_FREQ_FADD_R + RANDOMX_FREQ_FADD_M + RANDOMX_FREQ_FSUB_R + RANDOMX_FREQ_FSUB_M + \
 		RANDOMX_FREQ_FSCAL_R + RANDOMX_FREQ_FMUL_R + RANDOMX_FREQ_FDIV_M + RANDOMX_FREQ_FSQRT_R + RANDOMX_FREQ_CBRANCH + \
 		RANDOMX_FREQ_CFROUND + RANDOMX_FREQ_ISTORE + RANDOMX_FREQ_NOP;

src/configuration.h

@@ -1,5 +1,6 @@
 /*
 Copyright (c) 2018-2019, tevador <tevador@gmail.com>
+Copyright (c) 2019, Wownero Inc., a Monero Enterprise Alliance partner company
 
 All rights reserved.
 
@@ -38,7 +39,7 @@ OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 #define RANDOMX_ARGON_LANES        1
 
 //Argon2d salt
-#define RANDOMX_ARGON_SALT         "RandomX\x03"
+#define RANDOMX_ARGON_SALT         "RandomWOW\x01"
 
 //Number of random Cache accesses per Dataset item. Minimum is 2.
 #define RANDOMX_CACHE_ACCESSES     8
@@ -56,16 +57,16 @@ OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 #define RANDOMX_PROGRAM_SIZE       256
 
 //Number of iterations during VM execution.
-#define RANDOMX_PROGRAM_ITERATIONS 2048
+#define RANDOMX_PROGRAM_ITERATIONS 1024
 
 //Number of chained VM executions per hash.
-#define RANDOMX_PROGRAM_COUNT      8
+#define RANDOMX_PROGRAM_COUNT      16
 
 //Scratchpad L3 size in bytes. Must be a power of 2.
-#define RANDOMX_SCRATCHPAD_L3      2097152
+#define RANDOMX_SCRATCHPAD_L3      1048576
 
 //Scratchpad L2 size in bytes. Must be a power of two and less than or equal to RANDOMX_SCRATCHPAD_L3.
-#define RANDOMX_SCRATCHPAD_L2      262144
+#define RANDOMX_SCRATCHPAD_L2      131072
 
 //Scratchpad L1 size in bytes. Must be a power of two (minimum 64) and less than or equal to RANDOMX_SCRATCHPAD_L2.
 #define RANDOMX_SCRATCHPAD_L1      16384
@@ -82,7 +83,7 @@ Total sum of frequencies must be 256
 */
 
 //Integer instructions
-#define RANDOMX_FREQ_IADD_RS       16
+#define RANDOMX_FREQ_IADD_RS       25
 #define RANDOMX_FREQ_IADD_M         7
 #define RANDOMX_FREQ_ISUB_R        16
 #define RANDOMX_FREQ_ISUB_M         7
@@ -96,23 +97,23 @@ Total sum of frequencies must be 256
 #define RANDOMX_FREQ_INEG_R         2
 #define RANDOMX_FREQ_IXOR_R        15
 #define RANDOMX_FREQ_IXOR_M         5
-#define RANDOMX_FREQ_IROR_R         8
-#define RANDOMX_FREQ_IROL_R         2
+#define RANDOMX_FREQ_IROR_R        10
+#define RANDOMX_FREQ_IROL_R         0
 #define RANDOMX_FREQ_ISWAP_R        4
 
 //Floating point instructions
-#define RANDOMX_FREQ_FSWAP_R        4
-#define RANDOMX_FREQ_FADD_R        16
+#define RANDOMX_FREQ_FSWAP_R        8
+#define RANDOMX_FREQ_FADD_R        20
 #define RANDOMX_FREQ_FADD_M         5
-#define RANDOMX_FREQ_FSUB_R        16
+#define RANDOMX_FREQ_FSUB_R        20
 #define RANDOMX_FREQ_FSUB_M         5
 #define RANDOMX_FREQ_FSCAL_R        6
-#define RANDOMX_FREQ_FMUL_R        32
+#define RANDOMX_FREQ_FMUL_R        20
 #define RANDOMX_FREQ_FDIV_M         4
 #define RANDOMX_FREQ_FSQRT_R        6
 
 //Control instructions
-#define RANDOMX_FREQ_CBRANCH       25
+#define RANDOMX_FREQ_CBRANCH       16
 #define RANDOMX_FREQ_CFROUND        1
 
 //Store instruction

src/cpu.cpp

@@ -62,9 +62,13 @@ namespace randomx {
 			cpuid(info, 0x00000007);
 			avx2_ = (info[1] & (1 << 5)) != 0;
 		}
-#elif defined(__aarch64__) && defined(HWCAP_AES)
+#elif defined(__aarch64__)
+	#if defined(HWCAP_AES)
 		long hwcaps = getauxval(AT_HWCAP);
 		aes_ = (hwcaps & HWCAP_AES) != 0;
+	#elif defined(__APPLE__)
+		aes_ = true;
+	#endif
 #endif
 		//TODO POWER8 AES
 	}

src/jit_compiler.hpp

@@ -36,6 +36,6 @@ OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 #include "jit_compiler_fallback.hpp"
 #endif
 
-#if defined(__OpenBSD__) || defined(__NetBSD__)
+#if defined(__OpenBSD__) || defined(__NetBSD__) || (defined(__APPLE__) && defined(__aarch64__))
 #define RANDOMX_FORCE_SECURE
 #endif

src/jit_compiler_a64.cpp

@@ -95,6 +95,10 @@ JitCompilerA64::JitCompilerA64()
 {
 	memset(reg_changed_offset, 0, sizeof(reg_changed_offset));
 	memcpy(code, (void*) randomx_program_aarch64, CodeSize);
+
+#ifdef __GNUC__
+	__builtin___clear_cache(reinterpret_cast<char*>(code), reinterpret_cast<char*>(code + CodeSize));
+#endif
 }
 
 JitCompilerA64::~JitCompilerA64()

src/jit_compiler_a64_static.S

@@ -307,6 +307,9 @@ literal_v14: .fill 2,8,0
 literal_v15: .fill 2,8,0
 
 DECL(randomx_program_aarch64_vm_instructions_end):
+	# Calculate dataset pointer for dataset read
+	# Do it here to break false dependency from readReg2 and readReg3 (see next line)
+	lsr	x10, x9, 32
 
 	# mx ^= r[readReg2] ^ r[readReg3];
 	eor	x9, x9, x18
@@ -324,8 +327,6 @@ DECL(randomx_program_aarch64_cacheline_align_mask1):
 	# mx <-> ma
 	ror	x9, x9, 32
 
-	# Calculate dataset pointer for dataset read
-	mov	w10, w9
 DECL(randomx_program_aarch64_cacheline_align_mask2):
 	# Actual mask will be inserted by JIT compiler
 	and	x10, x10, 1

src/jit_compiler_x86.cpp

@@ -299,10 +299,6 @@ namespace randomx {
 			registerUsage[i] = -1;
 		}
 
-		codePos = ((uint8_t*)randomx_program_prologue_first_load) - ((uint8_t*)randomx_program_prologue);
-		code[codePos + sizeof(REX_XOR_RAX_R64)] = 0xc0 + pcfg.readReg0;
-		code[codePos + sizeof(REX_XOR_RAX_R64) * 2 + 1] = 0xc0 + pcfg.readReg1;
-
 		codePos = prologueSize;
 		memcpy(code + codePos - 48, &pcfg.eMask, sizeof(pcfg.eMask));
 		memcpy(code + codePos, codeLoopLoad, loopLoadSize);

src/jit_compiler_x86_static.S

@@ -40,7 +40,6 @@
 .global DECL(randomx_prefetch_scratchpad)
 .global DECL(randomx_prefetch_scratchpad_end)
 .global DECL(randomx_program_prologue)
-.global DECL(randomx_program_prologue_first_load)
 .global DECL(randomx_program_loop_begin)
 .global DECL(randomx_program_loop_load)
 .global DECL(randomx_program_start)
@@ -88,10 +87,6 @@ DECL(randomx_program_prologue):
 	movapd xmm13, xmmword ptr [mantissaMask+rip]
 	movapd xmm14, xmmword ptr [exp240+rip]
 	movapd xmm15, xmmword ptr [scaleMask+rip]
-
-DECL(randomx_program_prologue_first_load):
-	xor rax, r8
-	xor rax, r8
 	mov rdx, rax
 	and eax, RANDOMX_SCRATCHPAD_MASK
 	ror rdx, 32

src/jit_compiler_x86_static.asm

@@ -31,7 +31,6 @@ _RANDOMX_JITX86_STATIC SEGMENT PAGE READ EXECUTE
 PUBLIC randomx_prefetch_scratchpad
 PUBLIC randomx_prefetch_scratchpad_end
 PUBLIC randomx_program_prologue
-PUBLIC randomx_program_prologue_first_load
 PUBLIC randomx_program_loop_begin
 PUBLIC randomx_program_loop_load
 PUBLIC randomx_program_start
@@ -75,17 +74,12 @@ randomx_program_prologue PROC
 	movapd xmm13, xmmword ptr [mantissaMask]
 	movapd xmm14, xmmword ptr [exp240]
 	movapd xmm15, xmmword ptr [scaleMask]
-randomx_program_prologue ENDP
-
-randomx_program_prologue_first_load PROC
-	xor rax, r8
-	xor rax, r8
 	mov rdx, rax
 	and eax, RANDOMX_SCRATCHPAD_MASK
 	ror rdx, 32
 	and edx, RANDOMX_SCRATCHPAD_MASK
 	jmp randomx_program_loop_begin
-randomx_program_prologue_first_load ENDP
+randomx_program_prologue ENDP
 
 ALIGN 64
 	include asm/program_xmm_constants.inc
@@ -224,4 +218,4 @@ _RANDOMX_JITX86_STATIC ENDS
 
 ENDIF
 
-END
+END

src/jit_compiler_x86_static.hpp

@@ -32,7 +32,6 @@ extern "C" {
 	void randomx_prefetch_scratchpad();
 	void randomx_prefetch_scratchpad_end();
 	void randomx_program_prologue();
-	void randomx_program_prologue_first_load();
 	void randomx_program_loop_begin();
 	void randomx_program_loop_load();
 	void randomx_program_start();

src/virtual_memory.cpp

@@ -35,6 +35,11 @@ OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 #else
 #ifdef __APPLE__
 #include <mach/vm_statistics.h>
+#include <TargetConditionals.h>
+# if defined(__aarch64__) && TARGET_OS_OSX
+# define USE_PTHREAD_JIT_WP	1
+# include <pthread.h>
+# endif
 #endif
 #include <sys/types.h>
 #include <sys/mman.h>
@@ -99,9 +104,27 @@ void* allocMemoryPages(std::size_t bytes) {
 	#else
 		#define RESERVED_FLAGS 0
 	#endif
-	mem = mmap(nullptr, bytes, PAGE_READWRITE | RESERVED_FLAGS, MAP_ANONYMOUS | MAP_PRIVATE, -1, 0);
+	#ifdef __APPLE__
+		#include <TargetConditionals.h>
+		#ifdef TARGET_OS_OSX
+			#define MEXTRA MAP_JIT
+		#else
+			#define MEXTRA 0
+		#endif
+	#else
+		#define MEXTRA 0
+	#endif
+	#ifdef USE_PTHREAD_JIT_WP
+		#define PEXTRA	PROT_EXEC
+	#else
+		#define PEXTRA	0
+	#endif
+	mem = mmap(nullptr, bytes, PAGE_READWRITE | RESERVED_FLAGS | PEXTRA, MAP_ANONYMOUS | MAP_PRIVATE | MEXTRA, -1, 0);
 	if (mem == MAP_FAILED)
 		throw std::runtime_error("allocMemoryPages - mmap failed");
+#ifdef USE_PTHREAD_JIT_WP
+	pthread_jit_write_protect_np(false);
+#endif
 #endif
 	return mem;
 }
@@ -119,11 +142,19 @@ static inline void pageProtect(void* ptr, std::size_t bytes, int rules) {
 }
 
 void setPagesRW(void* ptr, std::size_t bytes) {
+#ifdef USE_PTHREAD_JIT_WP
+	pthread_jit_write_protect_np(false);
+#else
 	pageProtect(ptr, bytes, PAGE_READWRITE);
+#endif
 }
 
 void setPagesRX(void* ptr, std::size_t bytes) {
+#ifdef USE_PTHREAD_JIT_WP
+	pthread_jit_write_protect_np(true);
+#else
 	pageProtect(ptr, bytes, PAGE_EXECUTE_READ);
+#endif
 }
 
 void setPagesRWX(void* ptr, std::size_t bytes) {