cmd/5g, cmd/6g, cmd/8g: faster compilation

Replace linked list walk with memset.
This reduces CPU time taken by 'go install -a std' by ~10%.
Before:
real		user		sys
0m23.561s	0m16.625s	0m5.848s
0m23.766s	0m16.624s	0m5.846s
0m23.742s	0m16.621s	0m5.868s
after:
0m22.714s	0m14.858s	0m6.138s
0m22.644s	0m14.875s	0m6.120s
0m22.604s	0m14.854s	0m6.081s

R=golang-dev, r
CC=golang-dev
https://golang.org/cl/13084043

src/cmd/5g/peep.c

@@ -63,6 +63,8 @@ peep(Prog *firstp)
 	g = flowstart(firstp, sizeof(Flow));
 	if(g == nil)
 		return;
+	for(r=g->start, t=0; r!=nil; r=r->link, t++)
+		r->active = t;
 
 loop1:
 	if(debug['P'] && debug['v'])
@@ -343,6 +345,9 @@ gotit:
 	return 1;
 }
 
+static uchar *active;
+static int nactive;
+
 /*
  * The idea is to remove redundant copies.
  *	v1->v2	F=0
@@ -360,15 +365,17 @@ copyprop(Graph *g, Flow *r0)
 {
 	Prog *p;
 	Adr *v1, *v2;
-	Flow *r;
 
 	p = r0->prog;
 	v1 = &p->from;
 	v2 = &p->to;
 	if(copyas(v1, v2))
 		return 1;
-	for(r=g->start; r!=nil; r=r->link)
-		r->active = 0;
+	if(nactive < g->num) {
+		nactive = g->num;
+		active = realloc(active, g->num);
+	}
+ 	memset(active, 0, g->num);
 	return copy1(v1, v2, r0->s1, 0);
 }
 
@@ -378,12 +385,12 @@ copy1(Adr *v1, Adr *v2, Flow *r, int f)
 	int t;
 	Prog *p;
 
-	if(r->active) {
+	if(active[r->active]) {
 		if(debug['P'])
 			print("act set; return 1\n");
 		return 1;
 	}
-	r->active = 1;
+	active[r->active] = 1;
 	if(debug['P'])
 		print("copy %D->%D f=%d\n", v1, v2, f);
 	for(; r != nil; r = r->s1) {

src/cmd/6g/peep.c

@@ -92,6 +92,8 @@ peep(Prog *firstp)
 	g = flowstart(firstp, sizeof(Flow));
 	if(g == nil)
 		return;
+	for(r=g->start, t=0; r!=nil; r=r->link, t++)
+		r->active = t;
 
 	// byte, word arithmetic elimination.
 	elimshortmov(g);
@@ -626,6 +628,9 @@ gotit:
 	return 1;
 }
 
+static uchar *active;
+static int nactive;
+
 /*
  * The idea is to remove redundant copies.
  *	v1->v2	F=0
@@ -643,7 +648,6 @@ copyprop(Graph *g, Flow *r0)
 {
 	Prog *p;
 	Adr *v1, *v2;
-	Flow *r;
 
 	if(debug['P'] && debug['v'])
 		print("copyprop %P\n", r0->prog);
@@ -652,8 +656,11 @@ copyprop(Graph *g, Flow *r0)
 	v2 = &p->to;
 	if(copyas(v1, v2))
 		return 1;
-	for(r=g->start; r!=nil; r=r->link)
-		r->active = 0;
+	if(nactive < g->num) {
+		nactive = g->num;
+		active = realloc(active, g->num);
+	}
+ 	memset(active, 0, g->num);
 	return copy1(v1, v2, r0->s1, 0);
 }
 
@@ -663,12 +670,12 @@ copy1(Adr *v1, Adr *v2, Flow *r, int f)
 	int t;
 	Prog *p;
 
-	if(r->active) {
+	if(active[r->active]) {
 		if(debug['P'])
 			print("act set; return 1\n");
 		return 1;
 	}
-	r->active = 1;
+	active[r->active] = 1;
 	if(debug['P'])
 		print("copy %D->%D f=%d\n", v1, v2, f);
 	for(; r != nil; r = r->s1) {

src/cmd/8g/peep.c

@@ -91,6 +91,8 @@ peep(Prog *firstp)
 	g = flowstart(firstp, sizeof(Flow));
 	if(g == nil)
 		return;
+	for(r=g->start, t=0; r!=nil; r=r->link, t++)
+		r->active = t;
 
 	// byte, word arithmetic elimination.
 	elimshortmov(g);
@@ -424,6 +426,9 @@ gotit:
 	return 1;
 }
 
+static uchar *active;
+static int nactive;
+
 /*
  * The idea is to remove redundant copies.
  *	v1->v2	F=0
@@ -441,15 +446,17 @@ copyprop(Graph *g, Flow *r0)
 {
 	Prog *p;
 	Adr *v1, *v2;
-	Flow *r;
 
 	p = r0->prog;
 	v1 = &p->from;
 	v2 = &p->to;
 	if(copyas(v1, v2))
 		return 1;
-	for(r=g->start; r!=nil; r=r->link)
-		r->active = 0;
+	if(nactive < g->num) {
+		nactive = g->num;
+		active = realloc(active, g->num);
+	}
+ 	memset(active, 0, g->num);
 	return copy1(v1, v2, r0->s1, 0);
 }
 
@@ -459,12 +466,12 @@ copy1(Adr *v1, Adr *v2, Flow *r, int f)
 	int t;
 	Prog *p;
 
-	if(r->active) {
+	if(active[r->active]) {
 		if(debug['P'])
 			print("act set; return 1\n");
 		return 1;
 	}
-	r->active = 1;
+	active[r->active] = 1;
 	if(debug['P'])
 		print("copy %D->%D f=%d\n", v1, v2, f);
 	for(; r != nil; r = r->s1) {