Re: [PATCH 2/2] tracing: Add free_trace_iter_content() helper function

[Steven Rostedt <rostedt@goodmis.org>]

On Sat, 15 Jul 2023 09:42:01 -0400
Steven Rostedt <rostedt@goodmis.org> wrote:

> On Sat, 15 Jul 2023 13:15:32 +0800
> Zheng Yejian <zhengyejian1@huawei.com> wrote:
> 
> > > @@ -6709,7 +6723,12 @@ static int tracing_open_pipe(struct inode *inode, struct file *filp)
> > >   	}
> > >   
> > >   	trace_seq_init(&iter->seq);
> > > -	iter->trace = tr->current_trace;
> > > +
> > > +	iter->trace = kzalloc(sizeof(*iter->trace), GFP_KERNEL);
> > > +	if (!iter->trace)
> > > +		goto fail;  
> > 
> > Hi, Steve, 'ret' may need to be set before `goto fail`:
> >    ret = -ENOMEM;
> > 
> 
> As I mentioned to Masami, this hunk of the patch didn't belong.
> Something got mixed up in the commit. This patch even depends on the
> previous patch to remove the allocation completely.
> 

I know what I did now. I first started writing this patch and saw that
the iter->trace was inconsistent and in one place allocated a copy and
here it did not. I started to make the copy here, when I realized that
there was no reason to make that copy. Then I wrote the first patch to
remove the copying, but forgot to remove the copying I added in this
patch! :-p

-- Steve