runtime: fix windows build

Currently windows crashes because early allocs in schedinit
try to allocate tiny memory blocks, but m->p is not yet setup.
I've considered calling procresize(1) earlier in schedinit,
but this refactoring is better and must fix the issue as well.
Fixes #7218.

R=golang-codereviews, r
CC=golang-codereviews
https://golang.org/cl/54570045

src/pkg/runtime/malloc.goc

@@ -42,7 +42,6 @@ runtime·mallocgc(uintptr size, uintptr typ, uint32 flag)
 	MCacheList *l;
 	MLink *v;
 	byte *tiny;
-	P *p;
 
 	if(size == 0) {
 		// All 0-length allocations use this pointer.
@@ -93,10 +92,9 @@ runtime·mallocgc(uintptr size, uintptr typ, uint32 flag)
 			// the allocator reduces number of allocations by ~12% and
 			// reduces heap size by ~20%.
 
-			p = m->p;
-			tinysize = p->tinysize;
+			tinysize = c->tinysize;
 			if(size <= tinysize) {
-				tiny = p->tiny;
+				tiny = c->tiny;
 				// Align tiny pointer for required (conservative) alignment.
 				if((size&7) == 0)
 					tiny = (byte*)ROUND((uintptr)tiny, 8);
@@ -104,12 +102,12 @@ runtime·mallocgc(uintptr size, uintptr typ, uint32 flag)
 					tiny = (byte*)ROUND((uintptr)tiny, 4);
 				else if((size&1) == 0)
 					tiny = (byte*)ROUND((uintptr)tiny, 2);
-				size1 = size + (tiny - p->tiny);
+				size1 = size + (tiny - c->tiny);
 				if(size1 <= tinysize) {
 					// The object fits into existing tiny block.
 					v = (MLink*)tiny;
-					p->tiny += size1;
-					p->tinysize -= size1;
+					c->tiny += size1;
+					c->tinysize -= size1;
 					m->mallocing = 0;
 					m->locks--;
 					if(m->locks == 0 && g->preempt)  // restore the preemption request in case we've cleared it in newstack
@@ -129,8 +127,8 @@ runtime·mallocgc(uintptr size, uintptr typ, uint32 flag)
 			// See if we need to replace the existing tiny block with the new one
 			// based on amount of remaining free space.
 			if(TinySize-size > tinysize) {
-				p->tiny = (byte*)v + size;
-				p->tinysize = TinySize - size;
+				c->tiny = (byte*)v + size;
+				c->tinysize = TinySize - size;
 			}
 			size = TinySize;
 			goto done;

src/pkg/runtime/malloc.h

@@ -296,6 +296,10 @@ struct MCache
 	// so they are grouped here for better caching.
 	int32 next_sample;		// trigger heap sample after allocating this many bytes
 	intptr local_cachealloc;	// bytes allocated (or freed) from cache since last lock of heap
+	// Allocator cache for tiny objects w/o pointers.
+	// See "Tiny allocator" comment in malloc.goc.
+	byte*	tiny;
+	uintptr	tinysize;
 	// The rest is not accessed on every malloc.
 	MCacheList list[NumSizeClasses];
 	// Local allocator stats, flushed during GC.

src/pkg/runtime/mgc0.c

@@ -68,6 +68,7 @@ clearpools(void)
 {
 	void **pool, **next;
 	P *p, **pp;
+	MCache *c;
 	uintptr off;
 	int32 i;
 
@@ -86,8 +87,11 @@ clearpools(void)
 
 	for(pp=runtime·allp; p=*pp; pp++) {
 		// clear tinyalloc pool
-		p->tiny = nil;
-		p->tinysize = 0;
+		c = p->mcache;
+		if(c != nil) {
+			c->tiny = nil;
+			c->tinysize = 0;
+		}
 		// clear defer pools
 		for(i=0; i<nelem(p->deferpool); i++)
 			p->deferpool[i] = nil;

src/pkg/runtime/runtime.h

@@ -385,11 +385,6 @@ struct P
 	MCache*	mcache;
 	Defer*	deferpool[5];	// pool of available Defer structs of different sizes (see panic.c)
 
-	// Allocator cache for tiny objects w/o pointers.
-	// See "Tiny allocator" comment in malloc.goc.
-	byte*	tiny;
-	uintptr	tinysize;
-
 	// Cache of goroutine ids, amortizes accesses to runtime·sched.goidgen.
 	uint64	goidcache;
 	uint64	goidcacheend;