threads: Retrieve default stack size from __heap_base/__data_end (#350)

When compiling with `-z stack-size` flag, only the main thread's stack
size is set to the specified value and other threads use musl's default value.
That's inconsistent with LLD's `-Wl,-stack_size`.

I think we can make it similar to MUSL's behavior, where thread's stack
size can be set via `PT_GNU_STACK` program header (via `-Wl,-z,stack-size`
flag).

Configuring stack size through `pthread_attr_t` still work as expected and
overrides the defaults ([pthread_create.c](be1ffd6a9e/libc-top-half/musl/src/thread/pthread_create.c (L362)))
default settings.

expected/wasm32-wasi-pthread/undefined-symbols.txt

@@ -1,4 +1,5 @@
 __addtf3
+__data_end
 __divtf3
 __eqtf2
 __extenddftf2
@@ -9,6 +10,7 @@ __fixunstfsi
 __floatsitf
 __floatunsitf
 __getf2
+__global_base
 __gttf2
 __heap_base
 __imported_wasi_snapshot_preview1_args_get

libc-top-half/musl/src/env/__init_tls.c

@@ -16,6 +16,43 @@
 volatile int __thread_list_lock;
 
 #ifndef __wasilibc_unmodified_upstream
+
+/* These symbols are generated by wasm-ld. __stack_high/__stack_low
+ * symbols are only available in LLVM v16 and higher, therefore they're
+ * defined as weak symbols and if not available, __heap_base/__data_end
+ * is used instead.
+ *
+ * TODO: remove usage of __heap_base/__data_end for stack size calculation
+ * once we drop support for LLVM v15 and older.
+ */
+extern unsigned char __heap_base;
+extern unsigned char __data_end;
+extern unsigned char __global_base;
+extern weak unsigned char __stack_high;
+extern weak unsigned char __stack_low;
+
+static inline void setup_default_stack_size()
+{
+	ptrdiff_t stack_size;
+
+	if (&__stack_high)
+		stack_size = &__stack_high - &__stack_low;
+	else {
+		unsigned char *sp;
+		__asm__(
+			".globaltype __stack_pointer, i32\n"
+			"global.get __stack_pointer\n"
+			"local.set %0\n"
+			: "=r"(sp));
+		stack_size = sp > &__global_base ? &__heap_base - &__data_end : (ptrdiff_t)&__global_base;
+	}
+
+	if (stack_size > __default_stacksize)
+		__default_stacksize =
+			stack_size < DEFAULT_STACK_MAX ?
+			stack_size : DEFAULT_STACK_MAX;
+}
+
 void __wasi_init_tp() {
 	__init_tp((void *)__get_tp());
 }
@@ -31,6 +68,8 @@ int __init_tp(void *p)
 	if (!r) libc.can_do_threads = 1;
 	td->detach_state = DT_JOINABLE;
 	td->tid = __syscall(SYS_set_tid_address, &__thread_list_lock);
+#else
+	setup_default_stack_size();
 #endif
 	td->locale = &libc.global_locale;
 	td->robust_list.head = &td->robust_list.head;