Clean up abort.

* A dlmalloc usage error shouldn't call abort(3) because we want to
  cause a SIGSEGV by writing the address dlmalloc didn't like to an
  address the kernel won't like, so that debuggerd will dump the
  memory around the address that upset dlmalloc.

* Switch to the simpler FreeBSD/NetBSD style of registering stdio
  cleanup. Hopefully this will let us simplify more of the stdio
  implementation.

* Clear the stdio cleanup handler before we abort because of a dlmalloc
  corruption error. This fixes the reported bug, where we'd hang inside
  dlmalloc because the stdio cleanup reentered dlmalloc.

Bug: 9301265
Change-Id: Ief31b389455d6876e5a68f0f5429567d37277dbc

libc/stdio/findfp.c

@@ -171,10 +171,9 @@ f_prealloc(void)
 #endif
 
 /*
- * exit() and abort() call _cleanup() through the callback registered
- * with __atexit_register_cleanup(), set whenever we open or buffer a
- * file. This chicanery is done so that programs that do not use stdio
- * need not link it all in.
+ * exit() calls _cleanup() through *__cleanup, set whenever we
+ * open or buffer a file.  This chicanery is done so that programs
+ * that do not use stdio need not link it all in.
  *
  * The name `_cleanup' is, alas, fairly well known outside stdio.
  */
@@ -201,7 +200,7 @@ __sinit(void)
 		_FILEEXT_SETUP(usual+i, usualext+i);
 	}
 	/* make sure we clean up on exit */
-	__atexit_register_cleanup(_cleanup); /* conservative */
+	__cleanup = _cleanup; /* conservative */
 	__sdidinit = 1;
 out:
 	_THREAD_PRIVATE_MUTEX_UNLOCK(__sinit_mutex);