Re: [PATCH v4 4/7] perf diff: Use hists to manage basic blocks per symbol

[Jiri Olsa]

On Mon, Jun 24, 2019 at 10:49:25PM +0800, Jin, Yao wrote:
> 
> 
> On 6/24/2019 3:57 PM, Jiri Olsa wrote:
> > On Thu, Jun 20, 2019 at 10:36:39PM +0800, Jin Yao wrote:
> > 
> > SNIP
> > 
> > > +
> > > +static void *block_entry_zalloc(size_t size)
> > > +{
> > > +	return zalloc(size + sizeof(struct hist_entry));
> > > +}
> > > +
> > > +static void block_entry_free(void *he)
> > > +{
> > > +	struct block_info *bi = ((struct hist_entry *)he)->block_info;
> > > +
> > > +	block_info__put(bi);
> > > +	free(he);
> > > +}
> > > +
> > > +struct hist_entry_ops block_entry_ops = {
> > > +	.new    = block_entry_zalloc,
> > > +	.free   = block_entry_free,
> > > +};
> > 
> > hum, so there's already block_hist_ops moving that stuff into 'struct block_hist',
> > which is great, but why don't we have 'struct block_entry' in here? that would
> > keep the 'struct block_info'
> > 
> > thanks,
> > jirka
> > 
> 
> Hi Jiri,
> 
> If I define 'struct block_entry' as following and cast 'he' to 'struct
> block_entry' in some places, such as in block_cmp(), we can get the 'struct
> block_entry'.
> 
> struct block_entry {
> 	struct block_info bi;
> 	struct hist_entry he;
> };
> 
> But I don't know when I can set the 'bi' of 'struct block_entry'. Before or
> after calling hists__add_entry_xxx()? Before calling hists__add_entry_xxx(),
> we don't know the hist_entry. After calling hists__add_entry_xxx(), actually
> the hist_entry__cmp doesn't work (no bi ).
> 
> That's why I create block_info in hist_entry. Maybe I misunderstand what
> your suggested, correct me if I'm wrong.

ah you need the data already in place when calling hists__add_entry_block,
because of the block_cmp sorting function.. ok, it's the same tyhing we
do in c2c with mem_info.. I guess we can survive one pointer

I'll check the rest

thanks,
jirka