Clarifications/corrections in the specification/comments

Removed some unused code

doc/specs.md

@@ -86,7 +86,7 @@ and outputs a 256-bit result `R`.
 
 The algorithm consists of the following steps:
 
-1. The Dataset is initialized using the key value `K` (see chapter 6 for details).
+1. The Dataset is initialized using the key value `K` (see chapter 7 for details).
 1. 64-byte seed `S` is calculated as `S = Hash512(H)`.
 1. AesGenerator is initialized with state `S`.
 1. The Scratchpad is filled with `RANDOMX_SCRATCHPAD_L3` random bytes obtained from the AesGenerator.
@@ -383,8 +383,8 @@ The loop described below is repeated until the value of the `ic` register reache
 3. `spAddr1` is used to perform a 64-byte aligned read from Scratchpad level 3 (using mask from Table 4.2.1). Each floating point register `f0`-`f3` and `e0`-`e3` is initialized using an 8-byte value according to the conversion rules from chapters 4.3.1 and 4.3.2.
 4. The 256 instructions stored in the Program Buffer are executed.
 5. The `mx` register is XORed with the low 32 bits of registers `readReg2` and `readReg3` (see Table 4.5.3).
-6. A 64-byte memory block at address `datasetOffset + mx` is prefetched from the Dataset (this has no effect on the VM state).
+6. A 64-byte Dataset item at address `datasetOffset + mx % RANDOMX_DATASET_BASE_SIZE` is prefetched from the Dataset (it will be used during the next iteration).
-7. A 64-byte memory block at address `datasetOffset + ma` is loaded from the Dataset. The 64 bytes are XORed with all integer registers in order `r0`-`r7`.
+7. A 64-byte Dataset item at address `datasetOffset + ma % RANDOMX_DATASET_BASE_SIZE` is loaded from the Dataset. The 64 bytes are XORed with all integer registers in order `r0`-`r7`.
 8. The values of registers `mx` and `ma` are swapped.
 9. The values of all integer registers `r0`-`r7` are written to the Scratchpad (L3) at address `spAddr1` (64-byte aligned).
 10. Register `f0` is XORed with register `e0` and the result is stored in register `f0`. Register `f1` is XORed with register `e1` and the result is stored in register `f1`. Register `f2` is XORed with register `e2` and the result is stored in register `f2`. Register `f3` is XORed with register `e3` and the result is stored in register `f3`.
@@ -621,8 +621,8 @@ A register is considered as modified by an instruction in the following cases:
 There are 3 rules for the selection of the `creg` register, evaluated in this order:
 
 1. The register with the lowest value of `lastUsed` tag is selected.
-1. In case multiple registers have the same value of the `lastUsed` tag, the register with the lowest value of the `count` tag is selected.
+1. In case multiple registers have the same value of the `lastUsed` tag, the register with the lowest value of the `count` tag is selected from them.
-1. In case multiple registers have the same values of both `lastUsed` and `count` tags, a register with the lowest index is selected (`r0` before `r1` etc.).
+1. In case multiple registers have the same values of both `lastUsed` and `count` tags, the register with the lowest index is selected (`r0` before `r1` etc.) from them.
 
 Whenever a register is selected as the operand of a CBRANCH instruction, its `count` tag is increased by 1.
 

src/assembly_generator_x86.hpp

@@ -58,7 +58,6 @@ namespace randomx {
 		void h_ISMULH_R(Instruction&, int);
 		void h_ISMULH_M(Instruction&, int);
 		void h_IMUL_RCP(Instruction&, int);
-		void h_ISDIV_C(Instruction&, int);
 		void h_INEG_R(Instruction&, int);
 		void h_IXOR_R(Instruction&, int);
 		void h_IXOR_M(Instruction&, int);

src/jit_compiler_x86.cpp

@@ -72,7 +72,7 @@ namespace randomx {
 	REGISTER ALLOCATION:
 
 	; rax -> temporary
-	; rbx -> loop counter "lc"
+	; rbx -> iteration counter "ic"
 	; rcx -> temporary
 	; rdx -> temporary
 	; rsi -> scratchpad pointer
@@ -100,8 +100,8 @@ namespace randomx {
 	; xmm10 -> "a2"
 	; xmm11 -> "a3"
 	; xmm12 -> temporary
-	; xmm13 -> mantissa mask    = 0x000fffffffffffff000fffffffffffff
+	; xmm13 -> E 'and' mask = 0x00ffffffffffffff00ffffffffffffff
-	; xmm14 -> exponent 2**-240 = 0x30f00000000xxxxx30f00000000xxxxx
+	; xmm14 -> E 'or' mask  = 0x3*00000000******3*00000000******
 	; xmm15 -> scale mask   = 0x81f000000000000081f0000000000000
 
 	*/
@@ -444,7 +444,7 @@ namespace randomx {
 		emit32(instr.getModMem() ? ScratchpadL1Mask : ScratchpadL2Mask);
 	}
 
-	void JitCompilerX86::genAddressRegDst(Instruction& instr, bool align16 = false) {
+	void JitCompilerX86::genAddressRegDst(Instruction& instr) {
 		emit(LEA_32);
 		emitByte(0x80 + instr.dst);
 		if (instr.dst == RegisterNeedsSib) {
@@ -453,9 +453,7 @@ namespace randomx {
 		emit32(instr.getImm32());
 		emitByte(AND_EAX_I);
 		if (instr.getModCond() < StoreL3Condition) {
-			int32_t maskL1 = align16 ? ScratchpadL1Mask16 : ScratchpadL1Mask;
+			emit32(instr.getModMem() ? ScratchpadL1Mask : ScratchpadL2Mask);
-			int32_t maskL2 = align16 ? ScratchpadL2Mask16 : ScratchpadL2Mask;
-			emit32(instr.getModMem() ? maskL1 : maskL2);
 		}
 		else {
 			emit32(ScratchpadL3Mask);

src/jit_compiler_x86.hpp

@@ -65,7 +65,7 @@ namespace randomx {
 		void generateProgramPrologue(Program&, ProgramConfiguration&);
 		void generateProgramEpilogue(Program&);
 		void genAddressReg(Instruction&, bool);
-		void genAddressRegDst(Instruction&, bool);
+		void genAddressRegDst(Instruction&);
 		void genAddressImm(Instruction&);
 		void genSIB(int scale, int index, int base);
 

src/reciprocal.c

@@ -21,6 +21,7 @@ along with RandomX.  If not, see<http://www.gnu.org/licenses/>.
 
 /*
 	Calculates rcp = 2**x / divisor for highest integer x such that rcp < 2**64.
+	divisor must not be 0 or a power of 2
 
 	Equivalent x86 assembly (divisor in rcx):