Re: [PATCH 4/6] tracing: Change trace_seq to use separate buffer

[Frederic Weisbecker <fweisbec@gmail.com>]

On Tue, Jan 26, 2010 at 11:24:47AM +0800, Lai Jiangshan wrote:
> Frederic Weisbecker wrote:
> > On Tue, Jan 19, 2010 at 03:34:16PM +0800, Lai Jiangshan wrote:
> >> @@ -3124,6 +3126,8 @@ waitagain:
> >>  	if (cnt >= PAGE_SIZE)
> >>  		cnt = PAGE_SIZE - 1;
> >>  
> >> +	trace_seq_reset(&iter->seq);
> >> +
> > 
> > 
> > 
> > So we actually add a trace_seq_reset here.
> > This should have been in the first patch, which drops
> > the memset, and eventually modified here, just to avoid
> > breaking things in the middle of a patchset.
> > 
> > Things were already broken though before the memset dropping
> > patch though in other ways, so it's not that important I guess...
> > 
> > 
> 
> There is no trace_seq_reset() before this patch applied.
> trace_seq_init() in the first patch, has already reset it.



We have trace_seq_init() in the beginning of tracing_read_pipe()
but not in case we loop in waitagain.

In the first patch you've removed this:

	memset(&iter->seq, 0,
	       sizeof(struct trace_iterator) -
	       offsetof(struct trace_iterator, seq));

Which was a reset in waitagain.

And you re-add this reset in this patch, while it should have
been in the first one?

May be I miss something, that's not important really, things
were broken with the memset.