runtime: use new CNT_MASK in lfstack

This is for SPARC64, a 64-bit processor that uses all 64-bits
of virtual addresses.  The idea is to use the low order 3 bits
to at least get a small ABA counter.  That should work since
pointers are aligned.  The idea is for SPARC64 to set CNT_MASK
== 7, PTR_BITS == 0, PTR_MASK == 0xffffffffffffff8.

Also add uintptr casts to avoid GCC warnings.  The gccgo
runtime code is compiled with GCC, and GCC warns when casting
between a pointer and a type of a different size.

R=dvyukov
CC=golang-dev
https://golang.org/cl/7225043

src/pkg/runtime/lfstack.c

@@ -15,22 +15,23 @@
 # define PTR_BITS 32
 #endif
 #define PTR_MASK ((1ull<<PTR_BITS)-1)
+#define CNT_MASK (0ull-1)
 
 void
 runtime·lfstackpush(uint64 *head, LFNode *node)
 {
 	uint64 old, new;
 
-	if((uint64)node != ((uint64)node&PTR_MASK)) {
+	if((uintptr)node != ((uintptr)node&PTR_MASK)) {
 		runtime·printf("p=%p\n", node);
 		runtime·throw("runtime·lfstackpush: invalid pointer");
 	}
 
 	node->pushcnt++;
-	new = (uint64)node|(((uint64)node->pushcnt)<<PTR_BITS);
+	new = (uint64)(uintptr)node|(((uint64)node->pushcnt&CNT_MASK)<<PTR_BITS);
 	old = runtime·atomicload64(head);
 	for(;;) {
-		node->next = (LFNode*)(old&PTR_MASK);
+		node->next = (LFNode*)(uintptr)(old&PTR_MASK);
 		if(runtime·cas64(head, &old, new))
 			break;
 	}
@@ -46,11 +47,11 @@ runtime·lfstackpop(uint64 *head)
 	for(;;) {
 		if(old == 0)
 			return nil;
-		node = (LFNode*)(old&PTR_MASK);
+		node = (LFNode*)(uintptr)(old&PTR_MASK);
 		node2 = runtime·atomicloadp(&node->next);
 		new = 0;
 		if(node2 != nil)
-			new = (uint64)node2|(((uint64)node2->pushcnt)<<PTR_BITS);
+			new = (uint64)(uintptr)node2|(((uint64)node2->pushcnt&CNT_MASK)<<PTR_BITS);
 		if(runtime·cas64(head, &old, new))
 			return node;
 	}