Re: [PATCHv3 1/7] uprobe: Add support for session consumer

On Thu, Sep 12, 2024 at 06:35:39PM +0200, Oleg Nesterov wrote:

On 09/09, Jiri Olsa wrote:

quoted

 handle_uretprobe_chain(struct return_instance *ri, struct pt_regs *regs)
 {
+	struct return_consumer *ric = NULL;
 	struct uprobe *uprobe = ri->uprobe;
 	struct uprobe_consumer *uc;
-	int srcu_idx;
+	int srcu_idx, iter = 0;

 	srcu_idx = srcu_read_lock(&uprobes_srcu);
 	list_for_each_entry_srcu(uc, &uprobe->consumers, cons_node,
 				 srcu_read_lock_held(&uprobes_srcu)) {
+		/*
+		 * If we don't find return consumer, it means uprobe consumer
+		 * was added after we hit uprobe and return consumer did not
+		 * get registered in which case we call the ret_handler only
+		 * if it's not session consumer.
+		 */
+		ric = return_consumer_find(ri, &iter, uc->id);
+		if (!ric && uc->session)
+			continue;
 		if (uc->ret_handler)
-			uc->ret_handler(uc, ri->func, regs);
+			uc->ret_handler(uc, ri->func, regs, ric ? &ric->cookie : NULL);

So why do we need the new uc->session member and the uc->session above ?

If return_consumer_find() returns NULL, uc->ret_handler(..., NULL) can handle
this case itself?

I tried to explain that in the comment above.. we do not want to
execute session ret_handler at all in this case, because its entry
counterpart did not run

jirka

Oleg.