tal: simplify.

All the effort trying to keep the header size down to 2 pointers turns
out to be wasted.  In addition, getting the parent of a tal pointer is
now much faster.

./samba-allocs talloc.dump --tal-size:
Before:
	Virtual size = 9633792, RSS = 3952640
After:
	Virtual size = 9793536, RSS = 3948544

And we're much faster now, esp. on free.

./samba-allocs talloc.dump --tal:
Before:
	Tal time:                2718068ns
	Tal_free time:           3360258ns
	Single tal_free time:    1667412ns
After:
	Tal time:                2788287ns
	Tal_free time:           2290167ns
	Single tal_free time:    1566998ns
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>


Header from folded patch 'tal-fix-set-destroying-bit.patch':

tal: fix destroying bit usage.

Actually looking at a destroying contexts' parent didn't work, and a couple
of places didn't filter out the destroying bit.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

ccan/tal/_info

@@ -18,10 +18,10 @@
  * tal_free(X->name) would free X->name as expected, by tal_free(X) would
  * free X and X->name.
  *
- * With an overhead of approximately 2.1 pointers per object (vs. talloc's
- * 12 pointers), it's a little slower in freeing single objects, though
- * comparable for allocation and freeing whole object trees).  It does not
- * support talloc's references or failing destructors.
+ * With an overhead of approximately 4 pointers per object
+ * (vs. talloc's 12 pointers), it uses dynamic allocation for
+ * destructors and child lists, so those operations can fail.  It does
+ * not support talloc's references or failing destructors.
  *
  * Example:
  *	#include <stdio.h>
@@ -92,7 +92,6 @@ int main(int argc, char *argv[])
 
 	if (strcmp(argv[1], "depends") == 0) {
 		printf("ccan/compiler\n");
-		printf("ccan/hash\n");
 		printf("ccan/likely\n");
 		printf("ccan/list\n");
 		printf("ccan/str\n");

ccan/tal/benchmark/Makefile

@@ -6,8 +6,8 @@ LDLIBS=-lrt
 
 all: speed samba-allocs
 
-speed: speed.o tal.o talloc.o time.o hash.o list.o take.o
-samba-allocs: samba-allocs.o tal.o talloc.o time.o hash.o list.o take.o
+speed: speed.o tal.o talloc.o time.o list.o take.o
+samba-allocs: samba-allocs.o tal.o talloc.o time.o list.o take.o
 
 tal.o: ../tal.c
 	$(CC) $(CFLAGS) -c -o $@ $<
@@ -15,8 +15,6 @@ talloc.o: ../../talloc/talloc.c
 	$(CC) $(CFLAGS) -c -o $@ $<
 time.o: ../../time/time.c
 	$(CC) $(CFLAGS) -c -o $@ $<
-hash.o: ../../hash/hash.c
-	$(CC) $(CFLAGS) -c -o $@ $<
 list.o: ../../list/list.c
 	$(CC) $(CFLAGS) -c -o $@ $<
 take.o: ../../take/take.c

ccan/tal/test/run-array.c

@@ -40,7 +40,6 @@ int main(void)
 	strcpy(c[0], "hello");
 	tal_free(c[0]);
 	ok1(tal_first(parent) == NULL);
-
 	tal_free(parent);
 
 	return exit_status();

ccan/tal/test/run-iter.c

@@ -6,21 +6,14 @@
 
 int main(void)
 {
-	char *p[NUM], *iter;
+	char *p[NUM] = { NULL }, *iter;
 	int i;
 
 	plan_tests(NUM + 1 + NUM);
 
-	/* Create a random tree, but make sure we get multiple
-	 * top-level groups! */
+	/* Create a random tree */
 	for (i = 0; i < NUM; i++) {
-		p[i] = tal(NULL, char);
-		*p[i] = '0';
-		if (next_group(&null_parent.c.group) != &null_parent.c.group)
-			break;
-	}
-	for (i++; i < NUM; i++) {
-		p[i] = tal(p[rand() % i], char);
+		p[i] = tal(p[rand() % (i + 1)], char);
 		*p[i] = '0';
 	}
 

ccan/tal/test/run-steal.c

@@ -9,7 +9,7 @@ int main(void)
 
 	plan_tests(9);
 
-	p[0] = NULL;
+	p[0] = tal(NULL, char);
 	for (i = 1; i < 5; i++)
 		p[i] = tal(p[i-1], char);