Re: [RFC PATCH 5/4] tracing: Simplify the ftrace_event_field iteration in f_next/f_show

[Oleg Nesterov <oleg@redhat.com>]

On 07/17, Oleg Nesterov wrote:
>
> f_next() looks overcomplicated, and it is not strictly correct
> even if this doesn't matter.

Honestly, I do not know what the changelog can say to explain
this patch except "make the code simpler/cleaner".

To simplify the review, please see f_next/f_start with this patch
applied below. f_show() does field = list_entry(v, ...).

Probably makes sense anyway, I even tried to test it. So I would
not mind if you apply it as a separate cleanup ;)

Oleg.

static void *f_next(struct seq_file *m, void *v, loff_t *pos)
{
	struct ftrace_event_call *call = m->private;
	struct list_head *common_head = &ftrace_common_fields;
	struct list_head *head = trace_get_fields(call);
	struct list_head *node = v;

	(*pos)++;

	switch ((unsigned long)v) {
	case FORMAT_HEADER:
		node = common_head;
		break;

	case FORMAT_FIELD_SEPERATOR:
		node = head;
		break;

	case FORMAT_PRINTFMT:
		/* all done */
		return NULL;
	}

	node = node->prev;
	if (node == common_head)
		return (void *)FORMAT_FIELD_SEPERATOR;
	else if (node == head)
		return (void *)FORMAT_PRINTFMT;
	else
		return node;
}

static void *f_start(struct seq_file *m, loff_t *pos)
{
	void *p = (void *)FORMAT_HEADER;
	loff_t l = 0;

	while (p && l < *pos)
		p = f_next(m, p, &l);

	return p;
}