Re: [RFC PATCH v2 17/18] livepatch: change to a per-task consistency model

From
Josh Poimboeuf <hidden>
Date
2016-05-16 17:28:01
Also in
linux-s390, lkml
Reply to
Miroslav Benes, 2016-05-09
On Mon, May 09, 2016 at 11:41:37AM +0200, Miroslav Benes wrote:
quoted
+void klp_init_transition(struct klp_patch *patch, int state)
+{
+	struct task_struct *g, *task;
+	unsigned int cpu;
+	struct klp_object *obj;
+	struct klp_func *func;
+	int initial_state = !state;
+
+	klp_transition_patch = patch;
+
+	/*
+	 * If the patch can be applied or reverted immediately, skip the
+	 * per-task transitions.
+	 */
+	if (patch->immediate)
+		return;
+
+	/*
+	 * Initialize all tasks to the initial patch state to prepare them for
+	 * switching to the target state.
+	 */
+	read_lock(&tasklist_lock);
+	for_each_process_thread(g, task)
+		task->patch_state = initial_state;
+	read_unlock(&tasklist_lock);
+
+	/*
+	 * Ditto for the idle "swapper" tasks.
+	 */
+	get_online_cpus();
+	for_each_online_cpu(cpu)
+		idle_task(cpu)->patch_state = initial_state;
+	put_online_cpus();
+
+	/*
+	 * Ensure klp_ftrace_handler() sees the task->patch_state updates
+	 * before the func->transition updates.  Otherwise it could read an
+	 * out-of-date task state and pick the wrong function.
+	 */
+	smp_wmb();
+
+	/*
+	 * Set the func transition states so klp_ftrace_handler() will know to
+	 * switch to the transition logic.
+	 *
+	 * When patching, the funcs aren't yet in the func_stack and will be
+	 * made visible to the ftrace handler shortly by the calls to
+	 * klp_patch_object().
+	 *
+	 * When unpatching, the funcs are already in the func_stack and so are
+	 * already visible to the ftrace handler.
+	 */
+	klp_for_each_object(patch, obj)
+		klp_for_each_func(obj, func)
+			func->transition = true;
+
+	/*
+	 * Set the global target patch state which tasks will switch to.  This
+	 * has no effect until the TIF_PATCH_PENDING flags get set later.
+	 */
+	klp_target_state = state;
I am afraid there is a problem for (patch->immediate == true) patches. 
klp_target_state is not set for those and the comment is not entirely 
true, because klp_target_state has an effect in several places.
Ah, you're right.  I moved this assignment here for v2.  It was
originally done before the patch->immediate check.  If I remember
correctly, I moved it closer to the barrier for better readability (but
I created a bug in the process).
I guess we need to set klp_target_state even for immediate patches. Should 
we also initialize it with KLP_UNDEFINED and set it to KLP_UNDEFINED in 
klp_complete_transition()?
Yes, to both.

-- 
Josh
Thread (118 messages) 118 messages, 12 authors, last reply 2016-06-23
Keyboard shortcuts
hback out one level
jnext message in thread / next series-diff position
kprevious message in thread / previous series-diff position
ldrill in
Escclose help / fold thread tree
?toggle this help