Merge pull request #3013 from WojciechMula/simplify-asm

Simplify HUF_decompress4X2_usingDTable_internal_bmi2_asm_loop

lib/decompress/huf_decompress_amd64.S

@@ -427,41 +427,30 @@ HUF_decompress4X2_usingDTable_internal_bmi2_asm_loop:
     /* r15 = (ip0 - ilimit) / 7 */
     movq %rdx, %r15
 
-    movabsq $-3689348814741910323, %rdx
-    movq 8(%rsp), %rax /* rax = oend0 */
-    subq %op0,    %rax /* rax = oend0 - op0 */
-    mulq %rdx
-    shrq $3,      %rdx /* rdx = rax / 10 */
+    /* r15 = min(r15, min(oend0 - op0, oend1 - op1, oend2 - op2, oend3 - op3) / 10) */
+    movq 8(%rsp),  %rax /* rax = oend0 */
+    subq %op0,     %rax /* rax = oend0 - op0 */
+    movq 16(%rsp), %rdx /* rdx = oend1 */
+    subq %op1,     %rdx /* rdx = oend1 - op1 */
 
-    /* r15 = min(%rdx, %r15) */
-    cmpq  %rdx, %r15
-    cmova %rdx, %r15
+    cmpq  %rax,    %rdx
+    cmova %rax,    %rdx /* rdx = min(%rdx, %rax) */
 
-    movabsq $-3689348814741910323, %rdx
-    movq 16(%rsp), %rax /* rax = oend1 */
-    subq %op1,     %rax /* rax = oend1 - op1 */
-    mulq %rdx
-    shrq $3,       %rdx /* rdx = rax / 10 */
-
-    /* r15 = min(%rdx, %r15) */
-    cmpq  %rdx, %r15
-    cmova %rdx, %r15
-
-    movabsq $-3689348814741910323, %rdx
     movq 24(%rsp), %rax /* rax = oend2 */
     subq %op2,     %rax /* rax = oend2 - op2 */
-    mulq %rdx
-    shrq $3,       %rdx /* rdx = rax / 10 */
 
-    /* r15 = min(%rdx, %r15) */
-    cmpq  %rdx, %r15
-    cmova %rdx, %r15
+    cmpq  %rax,    %rdx
+    cmova %rax,    %rdx /* rdx = min(%rdx, %rax) */
 
-    movabsq $-3689348814741910323, %rdx
     movq 32(%rsp), %rax /* rax = oend3 */
     subq %op3,     %rax /* rax = oend3 - op3 */
+
+    cmpq  %rax,    %rdx
+    cmova %rax,    %rdx /* rdx = min(%rdx, %rax) */
+
+    movabsq $-3689348814741910323, %rax
     mulq %rdx
-    shrq $3,       %rdx /* rdx = rax / 10 */
+    shrq $3,       %rdx /* rdx = rdx / 10 */
 
     /* r15 = min(%rdx, %r15) */
     cmpq  %rdx, %r15