diff --git a/headers/private/kernel/arch/thread.h b/headers/private/kernel/arch/thread.h
index 225f001c9b..8f11286d11 100644
--- a/headers/private/kernel/arch/thread.h
+++ b/headers/private/kernel/arch/thread.h
@@ -23,7 +23,7 @@ status_t arch_thread_init_tls(struct thread *thread);
 void arch_thread_context_switch(struct thread *t_from, struct thread *t_to);
 status_t arch_thread_init_kthread_stack(struct thread *t, int (*start_func)(void), void (*entry_func)(void), void (*exit_func)(void));
 void arch_thread_dump_info(void *info);
-void arch_thread_enter_uspace(struct thread *t, addr_t entry, void *args1, void *args2);
+status_t arch_thread_enter_userspace(struct thread *t, addr_t entry, void *args1, void *args2);
 void arch_thread_switch_kstack_and_call(struct thread *t, addr_t new_kstack, void (*func)(void *), void *arg);
 
 // ToDo: doing this this way is an ugly hack - please fix me!
diff --git a/headers/private/kernel/arch/x86/arch_cpu.h b/headers/private/kernel/arch/x86/arch_cpu.h
index 46fd99b4fe..ff70e45ab7 100644
--- a/headers/private/kernel/arch/x86/arch_cpu.h
+++ b/headers/private/kernel/arch/x86/arch_cpu.h
@@ -1,5 +1,5 @@
 /*
- * Copyright 2002-2006, Axel Dörfler, axeld@pinc-software.de. All rights reserved.
+ * Copyright 2002-2007, Axel Dörfler, axeld@pinc-software.de. All rights reserved.
  * Distributed under the terms of the MIT License.
  *
  * Copyright 2001-2002, Travis Geiselbrecht. All rights reserved.
@@ -100,7 +100,9 @@ struct arch_thread;
 
 void __x86_setup_system_time(uint32 cv_factor);
 void i386_context_switch(struct arch_thread *old_state, struct arch_thread *new_state, addr_t new_pgdir);
-void i386_enter_uspace(addr_t entry, void *args1, void *args2, addr_t ustack_top);
+void x86_userspace_thread_exit(void);
+void x86_end_userspace_thread_exit(void);
+void x86_enter_userspace(addr_t entry, void *args1, void *args2, addr_t ustack_top);
 void i386_set_tss_and_kstack(addr_t kstack);
 void i386_switch_stack_and_call(addr_t stack, void (*func)(void *), void *arg);
 void i386_swap_pgdir(addr_t new_pgdir);
diff --git a/src/system/kernel/arch/ppc/arch_thread.c b/src/system/kernel/arch/ppc/arch_thread.c
index 903b1c37bf..a416a65315 100644
--- a/src/system/kernel/arch/ppc/arch_thread.c
+++ b/src/system/kernel/arch/ppc/arch_thread.c
@@ -206,10 +206,11 @@ arch_thread_dump_info(void *info)
 }
 
 
-void
-arch_thread_enter_uspace(struct thread *thread, addr_t entry, void *arg1, void *arg2)
+status_t
+arch_thread_enter_userspace(struct thread *thread, addr_t entry, void *arg1, void *arg2)
 {
 	panic("arch_thread_enter_uspace(): not yet implemented\n");
+	return B_ERROR;
 }
 
 
diff --git a/src/system/kernel/arch/x86/arch_thread.c b/src/system/kernel/arch/x86/arch_thread.c
index 2f115c118f..e7cead4791 100644
--- a/src/system/kernel/arch/x86/arch_thread.c
+++ b/src/system/kernel/arch/x86/arch_thread.c
@@ -306,18 +306,22 @@ arch_thread_dump_info(void *info)
 /** Sets up initial thread context and enters user space
  */
 
-void
-arch_thread_enter_uspace(struct thread *t, addr_t entry, void *args1, void *args2)
+status_t
+arch_thread_enter_userspace(struct thread *t, addr_t entry, void *args1, void *args2)
 {
-	addr_t ustack_top = t->user_stack_base + t->user_stack_size;
+	addr_t stackTop = t->user_stack_base + t->user_stack_size;
+	uint32 codeSize = (addr_t)x86_end_userspace_thread_exit
+		- (addr_t)x86_userspace_thread_exit;
 
 	TRACE(("arch_thread_enter_uspace: entry 0x%lx, args %p %p, ustack_top 0x%lx\n",
-		entry, args1, args2, ustack_top));
+		entry, args1, args2, stackTop));
 
-	// access the new stack to make sure the memory page is present
-	// while interrupts are disabled.
-	// XXX does this belong there, should caller take care of it?
-	*(uint32 *)(ustack_top - 8) = 0;
+	// copy the little stub that calls exit_thread() when the thread entry
+	// function returns
+	stackTop -= codeSize;
+
+	if (user_memcpy((void *)stackTop, x86_userspace_thread_exit, codeSize) < B_OK)
+		return B_BAD_ADDRESS;
 
 	disable_interrupts();
 
@@ -326,7 +330,10 @@ arch_thread_enter_uspace(struct thread *t, addr_t entry, void *args1, void *args
 	// set the CPU dependent GDT entry for TLS
 	set_tls_context(t);
 
-	i386_enter_uspace(entry, args1, args2, ustack_top - 4);
+	x86_enter_userspace(entry, args1, args2, stackTop);
+
+	return B_OK;
+		// never gets here
 }
 
 
diff --git a/src/system/kernel/arch/x86/arch_x86.S b/src/system/kernel/arch/x86/arch_x86.S
index 695631f970..ef7741a1dd 100644
--- a/src/system/kernel/arch/x86/arch_x86.S
+++ b/src/system/kernel/arch/x86/arch_x86.S
@@ -1,5 +1,5 @@
 /*
- * Copyright 2003-2005, Axel Dörfler, axeld@pinc-software.de.
+ * Copyright 2003-2007, Axel Dörfler, axeld@pinc-software.de.
  * Distributed under the terms of the MIT License.
  *
  * Copyright 2001, Travis Geiselbrecht. All rights reserved.
@@ -126,20 +126,20 @@ FUNCTION(i386_swap_pgdir):
 	movl	%eax,%cr3
 	ret
 
-/* thread exit stub */
+/* thread exit stub - is copied to the userspace stack in arch_thread_enter_uspace() */
 	.align 4
-i386_uspace_exit_stub:
+FUNCTION(x86_userspace_thread_exit):
 	pushl	%eax
 	movl	$1, %ecx
-	lea	(%esp), %edx
+	lea		(%esp), %edx
 	movl	$SYSCALL_EXIT_THREAD, %eax;
-	int	$99
+	int		$99
 	.align 4
-i386_uspace_exit_stub_end:
+FUNCTION(x86_end_userspace_thread_exit):
 
 
-/* void i386_enter_uspace(addr entry, void *args1, void *args2, addr ustack_top); */
-FUNCTION(i386_enter_uspace):
+/* void x86_enter_userspace(addr entry, void *args1, void *args2, addr stackTop); */
+FUNCTION(x86_enter_userspace):
 	movl	4(%esp),%eax	// get entry point
 	movl	8(%esp),%edx	// get arguments
 	movl	12(%esp),%edi
@@ -150,16 +150,6 @@ FUNCTION(i386_enter_uspace):
 	//movw	$0x33 + cpu_num,%fs	-> fs points to the TLS storage (CPU dependent segment)
 	movw	%cx,%gs
 
-	// copy exit stub to stack
-	movl	$i386_uspace_exit_stub_end, %esi
-_copy_more:
-	lea	-4(%esi), %esi
-	lea	-4(%ebx), %ebx
-	mov	(%esi), %ecx
-	mov	%ecx, (%ebx)
-	cmp	$i386_uspace_exit_stub, %esi
-	jg	_copy_more
-
 	// push the args onto the user stack
 	movl	%edi,-4(%ebx)	// args1
 	movl	%edx,-8(%ebx)	// args2
diff --git a/src/system/kernel/team.c b/src/system/kernel/team.c
index 5ac18a2fc3..e250329fe1 100644
--- a/src/system/kernel/team.c
+++ b/src/system/kernel/team.c
@@ -819,10 +819,8 @@ team_create_thread_start(void *args)
 	team->state = TEAM_STATE_NORMAL;
 
 	// jump to the entry point in user space
-	arch_thread_enter_uspace(t, entry, uspa, NULL);
-
-	// never gets here
-	return B_OK;
+	return arch_thread_enter_userspace(t, entry, uspa, NULL);
+		// only returns in case of error
 }
 
 
diff --git a/src/system/kernel/thread.c b/src/system/kernel/thread.c
index 1d7ff170ed..df948c6524 100644
--- a/src/system/kernel/thread.c
+++ b/src/system/kernel/thread.c
@@ -309,9 +309,10 @@ _create_user_thread_kentry(void)
 	thread_at_kernel_exit();
 
 	// jump to the entry point in user space
-	arch_thread_enter_uspace(thread, (addr_t)thread->entry, thread->args1, thread->args2);
+	arch_thread_enter_userspace(thread, (addr_t)thread->entry,
+		thread->args1, thread->args2);
 
-	// never get here
+	// only get here if the above call fails
 	return 0;
 }
 
@@ -462,14 +463,12 @@ create_thread(const char *name, team_id teamID, thread_entry_func entry,
 				(void **)&thread->user_stack_base, B_BASE_ADDRESS,
 				thread->user_stack_size + TLS_SIZE, B_NO_LOCK,
 				B_READ_AREA | B_WRITE_AREA | B_STACK_AREA);
-		if (thread->user_stack_area < 0) {
-			// great, we have a fully running thread without a stack
-			dprintf("create_thread: unable to create user stack!\n");
+		if (thread->user_stack_area < B_OK
+			|| arch_thread_init_tls(thread) < B_OK) {
+			// great, we have a fully running thread without a (usable) stack
+			dprintf("create_thread: unable to create proper user stack!\n");
 			status = thread->user_stack_area;
 			kill_thread(thread->id);
-		} else {
-			// now that the TLS area is allocated, initialize TLS
-			arch_thread_init_tls(thread);
 		}
 
 		// copy the user entry over to the args field in the thread struct
@@ -603,8 +602,8 @@ dump_thread_info(int argc, char **argv)
 	}
 
 	if (argc == 1) {
-		name = NULL;
-		id = thread_get_current_thread()->id;
+		_dump_thread_info(thread_get_current_thread());
+		return 0;
 	} else {
 		name = argv[1];
 		id = strtoul(argv[1], NULL, 0);