jbuf: allocate mutex and lock also jbuf_debug()

[cspiel1]

For the RX thread baresip/baresip#2445 jbuf_debug() should also be thread safe.

[cspiel1]

Now I have a better solution. jbuf_flush() should check if the lock is NULL.

[sreimers]

Now I have a better solution. jbuf_flush() should check if the lock is NULL.

Thats not very reliable. Better fix allocation, so you can rely on mutex. So we handle this on other places too.

[cspiel1]

Now I have a better solution. jbuf_flush() should check if the lock is NULL.

Thats not very reliable. Better fix allocation, so you can rely on mutex. So we handle this on other places too.

Okay, coming soon.

[cspiel1]

How do you like the variant how jbuf_debug() was made thread safe, with short mutex lock phase? Any further ideas?

	struct mbuf *mb = mbuf_alloc(512);
	struct re_printf pfh = {
		vprintf_handler,
		mb
	};

[alfredh]

I think it is better to have straightforward and readable code.

I dont think that you get much performance benefit with this approach.

In general, clean and maintainable code should be preferred over performance.

[cspiel1]

@sreimers once wrote that printing to stdout may block sometimes.

I don't want to print to stdout while the mutex is locked. But I agree also with you and there might be a simpler solution with an mbuf. A commit will follow.