Re: [PATCH v2] printk: add console output tracing

[Johannes Berg]

On Fri, 2011-11-18 at 19:54 +0100, Frederic Weisbecker wrote:
> On Fri, Nov 18, 2011 at 07:46:15PM +0100, Johannes Berg wrote:
> > On Fri, 2011-11-18 at 19:44 +0100, Frederic Weisbecker wrote:
> > 
> > > > +TRACE_EVENT_CONDITION(console,
> > > > +	TP_PROTO(const char *log_buf, unsigned start, unsigned end,
> > > > +		 unsigned log_buf_len),
> > > > +
> > > > +	TP_ARGS(log_buf, start, end, log_buf_len),
> > > > +
> > > > +	TP_CONDITION(start != end),
> > > > +
> > > > +	TP_STRUCT__entry(
> > > > +		__dynamic_array(char, msg,
> > > > +				((end - start + log_buf_len) &
> > > > +				 (log_buf_len - 1)) + 1)
> > > 
> > > Is all that care about log_buf_len necessary? It seems that
> > > printk ensures that log_end - con_start never exceeds log_buf_len,
> > > looking at emit_log_char()
> > 
> > I think it is. The buffer can wrap around so in that case end < start,
> > which just end-start won't handle here.
> 
> Even if it wraps, end - start should always give a positive result.
> 
> We have that check in call_console_drivers():
> 
> 	BUG_ON(((int)(start - end)) > 0)

That's .. confusing, start - end > 0 <=> start > end ??

Also, call_console_drivers() doesn't pass this start/end through to
_call_console_drivers(), it has loops and stuff there...

In any case -- feel free to clean it all up, I basically copied the
logic from _call_console_drivers assuming it was needed. I don't really
want to know about the printk details :-)

johannes

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/