Add PopCount compiler intrinsic. Remove BCount

Remove set_bits_table

src/Compiler/CInit.CC

@@ -199,6 +199,7 @@ CIntermediateStruct intermediate_code_table[IC_ICS_NUM]={
   {IS_1_ARG,1,IST_NULL,TRUE,FALSE,0,0,0,"COS"},
   {IS_1_ARG,1,IST_NULL,TRUE,FALSE,0,0,0,"TAN"},
   {IS_1_ARG,1,IST_NULL,TRUE,FALSE,0,0,0,"ATAN"},
+  {IS_1_ARG,1,IST_NULL,FALSE,TRUE,0,0,0,"POPCNT"},
 };
 
 U0 CompLoadDefines()

src/Compiler/CompilerA.HH

@@ -116,7 +116,7 @@ class CIntermediateStruct
 #define IC_JMP			0x51
 #define IC_SUB_CALL		0x52
 #define IC_SWITCH		0x53
-#define IC_NOBOUND_SWITCH		0x54
+#define IC_NOBOUND_SWITCH	0x54
 
 #define IC_ADD_RSP		0x55
 #define IC_ADD_RSP1		0x56
@@ -171,7 +171,7 @@ class CIntermediateStruct
 
 #define IC_QUEUE_INIT		0x80
 #define IC_QUEUE_INSERT		0x81
-#define IC_QUEUE_INSERT_REV		0x82
+#define IC_QUEUE_INSERT_REV	0x82
 #define IC_QUEUE_REMOVE		0x83
 
 #define IC_STRLEN		0x84
@@ -234,7 +234,8 @@ class CIntermediateStruct
 #define IC_COS			0xB6
 #define IC_TAN			0xB7
 #define IC_ATAN			0xB8
-#define IC_ICS_NUM		0xB9
+#define IC_POPCNT		0xB9
+#define IC_ICS_NUM		0xBA
 
 #define KW_INCLUDE	0
 #define KW_DEFINE	1

src/Compiler/OptPass012.CC

@@ -931,6 +931,17 @@ args.
 	  OptSetNOP1(tmpi);		//BSR
 	}
 	break;
+      case IC_POPCNT:
+	if (tmpi1->ic_code==IC_IMM_I64) {
+	  tmpi1->ic_data = PopCount(tmpi1->ic_data);
+	  tmpi_push = tmpi1;
+	  OptSetNOP1(OptLag(tmpi1));	//CALL_START
+	  tmpi2 = OptLead1(tmpi);
+	  tmpi1->ic_flags |= tmpi->ic_flags | tmpi2->ic_flags;
+	  OptSetNOP1(tmpi2);		//CALL_END
+	  OptSetNOP1(tmpi);		//POPCNT
+	}
+        break;
       case IC_LBTS:
       case IC_LBTR:
       case IC_LBTC:

src/Compiler/OptPass3.CC

@@ -439,6 +439,7 @@ here1:
 	case IC_LBTC:
 	case IC_BSF:
 	case IC_BSR:
+        case IC_POPCNT:
 	case IC_SIGN_I64:
 	case IC_TOUPPER:
 	case IC_TO_I64:

src/Compiler/OptPass4.CC

@@ -619,6 +619,7 @@ p4_sib:
 	case IC_ADDR_IMPORT:
 	case IC_BSF:
 	case IC_BSR:
+        case IC_POPCNT:
 	case IC_SIGN_I64:
 	case IC_TOUPPER:
 	case IC_TO_I64:

src/Compiler/OptPass789A.CC

@@ -817,6 +817,12 @@ cc->pass==10 is final pass, code is placed into buf.
 	  ICU16(tmpi,0x0375);
 	  ICU24(tmpi,0xD0F748);
 	  break;
+        case IC_POPCNT:
+	  ICMov(tmpi,MDF_REG+RT_I64,REG_RAX,0,
+		tmpi->arg1.type,tmpi->arg1.reg,tmpi->arg1.disp,rip2);
+	  ICU16(tmpi, 0xF348);
+          ICU32(tmpi, 0xC0B80F48);
+          break;
 	case IC_SIGN_I64:
 	  ICMov(tmpi,MDF_REG+RT_I64,REG_RAX,0,
 		tmpi->arg1.type,tmpi->arg1.reg,tmpi->arg1.disp,rip2);

src/Doc/Bit.DD

@@ -9,6 +9,6 @@ $FG,2$BEqual$FG$:	Set bit to value.
 $FG$Bit operations are "atomic", no interrupt between the reading and writing the bit, important when multitasking.  For multicore use "locked" forms.$FG$
 
 $FG,5$These don't take a pointer, but the actual field.$FG$
-$FG,2$Bsf$FG$:	Bit Scan Fwd (Pos of first low one bit or -1)
-$FG,2$Bsr$FG$:	Bit Scan Rev (Pos of first high one bit or -1)
-$FG,2$BCount$FG$:	Bit Count (Count of set bits)
+$FG,2$Bsf$FG$:	Bit Scan Forward (Pos of first low one bit or -1)
+$FG,2$Bsr$FG$:	Bit Scan Reverse (Pos of first high one bit or -1)
+$FG,2$PopCount$FG$:Population Count (Count of set bits)

src/Kernel/KGlobals.CC

@@ -9,8 +9,7 @@ CTask *zenith_task;
 I64 sys_num_spawned_tasks;
 
 CTask *sys_winmgr_task,*sys_task_being_screen_updated;
-U8 *rev_bits_table, //Table with U8 bits revd
-   *set_bits_table; //Table with count of set bits in a U8
+U8 *rev_bits_table; //Table with U8 bits reversed
 CDate local_time_offset;
 F64	*pow10_I64,
 	sys_os_version=0.10;
@@ -25,7 +24,7 @@ CGridGlobals	mouse_grid; //See $LK,"::/Demo/Graphics/Grid.CC"$.
 CMouseStateGlobals	mouse,mouse_last;
 CKbdStateGlobals	kbd;
 CKeyDevGlobals	keydev;
-CMouseHardStateGlobals mouse_hard,mouse_hard_last;
+CMouseHardStateGlobals mouse_hard,mouse_hard_last;
 CScreenCastGlobals	screencast;
 CTextGlobals	text;
 

src/Kernel/KMain.CC

@@ -26,11 +26,9 @@ U0 SysGlobalsInit()
 
   DebugMode(ON);
   rev_bits_table=CAlloc(256);
-  set_bits_table=CAlloc(256);
   for (i=0;i<256;i++)
     for (j=0;j<8;j++) {
       if (Bt(&i,7-j))	LBts(rev_bits_table+i,j);
-      if (Bt(&i,j))	set_bits_table[i]++;
     }
 
   ext=CAlloc(EXT_EXTS_NUM*sizeof(U8 *));

src/Kernel/KMisc.CC

@@ -4,15 +4,7 @@ U0 PortNop()
   InU8(0x21);
 }
 
-I64 BCount(I64 d)
-{//Count set bits in I64.
-  I64 res=0,i;
-  for (i=0;i<8;i++)
-    res+=set_bits_table[d.u8[i]];
-  return res;
-}
-
-U0 IntCore0TimerHandler(CTask *)
+U0 IntCore0TimerHandler(CTask *)
 {//Called from $LK,"IntCore0TimerHandler",A="FF:::/Kernel/KInterrupts.CC,IntCore0TimerHandler"$
   I64 i;
   if (mp_count>1)

src/Kernel/KernelB.HH

@@ -6,32 +6,20 @@ $LK,"KernelB",A="FF:::/StartOS.CC,KernelB"$ StartOS.CC
 
 #help_index "Bit"
 #help_file "::/Doc/Bit"
-public _intern IC_BSF I64 Bsf(
-	I64 bit_field_val);//Scan forward from lowest for 1st set. -1 if not found.
-public _intern IC_BSR I64 Bsr(
-	I64 bit_field_val);//Scan rev from highest for 1st set. -1 if not found.
-public _intern IC_BT Bool Bt(
-	U8 *bit_field,I64 bit); //Bit test.
-public _intern IC_BTC Bool Btc(
-	U8 *bit_field,I64 bit); //Bit test and complement (same as xor with 1).
-public _intern IC_BTR Bool Btr(
-	U8 *bit_field,I64 bit); //Bit test and reset to zero.
-public _intern IC_BTS Bool Bts(
-	U8 *bit_field,I64 bit); //Bit test and set to one.
-public _intern IC_LBTC Bool LBtc(
-	U8 *bit_field,I64 bit); //Locked bit test and complement (xor with 1).
-public _intern IC_LBTR Bool LBtr(
-	U8 *bit_field,I64 bit); //Locked bit test and reset to zero.
-public _intern IC_LBTS Bool LBts(
-	U8 *bit_field,I64 bit); //Locked bit test and set to one.
-public _extern _BEQUAL Bool BEqual(
-	U8 *bit_field,I64 bit,Bool val);//Set bit equ to val.
-public _extern _BIT_FIELD_EXT_U32 U32 BFieldExtU32(
-	U8 *bit_field,I64 bit,I64 size); //Extract U32 from bit field.
-public _extern _BIT_FIELD_OR_U32 U0 BFieldOrU32(
-	U8 *bit_field,I64 bit,U32 pattern); //Or U32 into bit field.
-public _extern _LBEQUAL Bool LBEqual(
-	U8 *bit_field,I64 bit,Bool val); //Locked Set bit equ to val.
+public _intern IC_BSF I64 Bsf(I64 bit_field_val);//Scan forward from lowest for 1st set. -1 if not found.
+public _intern IC_BSR I64 Bsr(I64 bit_field_val);//Scan rev from highest for 1st set. -1 if not found.
+public _intern IC_BT Bool Bt(U8 *bit_field,I64 bit); //Bit test.
+public _intern IC_BTC Bool Btc(U8 *bit_field,I64 bit); //Bit test and complement (same as xor with 1).
+public _intern IC_BTR Bool Btr(U8 *bit_field,I64 bit); //Bit test and reset to zero.
+public _intern IC_BTS Bool Bts(U8 *bit_field,I64 bit); //Bit test and set to one.
+public _intern IC_LBTC Bool LBtc(U8 *bit_field,I64 bit); //Locked bit test and complement (xor with 1).
+public _intern IC_LBTR Bool LBtr(U8 *bit_field,I64 bit); //Locked bit test and reset to zero.
+public _intern IC_LBTS Bool LBts(U8 *bit_field,I64 bit); //Locked bit test and set to one.
+public _intern IC_POPCNT I64 PopCount(I64 bit_field_val);
+public _extern _BEQUAL Bool BEqual(U8 *bit_field,I64 bit,Bool val);//Set bit equ to val.
+public _extern _BIT_FIELD_EXT_U32 U32 BFieldExtU32(U8 *bit_field,I64 bit,I64 size); //Extract U32 from bit field.
+public _extern _BIT_FIELD_OR_U32 U0 BFieldOrU32(U8 *bit_field,I64 bit,U32 pattern); //Or U32 into bit field.
+public _extern _LBEQUAL Bool LBEqual(U8 *bit_field,I64 bit,Bool val); //Locked Set bit equ to val.
 
 #help_index "Boot"
 _extern MEM_BOOT_BASE U32 mem_boot_base;

src/Kernel/KernelC.HH

@@ -6,8 +6,7 @@ public extern CAutoCompleteGlobals ac;
 public extern CAutoCompleteDictGlobals acd;
 
 #help_index "Bit"
-public extern I64 BCount(I64 d);
-public extern U8 *rev_bits_table,*set_bits_table;
+public extern U8 *rev_bits_table;
 
 #help_index "Boot"
 #help_file "::/Doc/Boot"