Re: [PATCH 1/2] SDT markers listing by perf

[Hemant <hkshaw at linux dot vnet dot ibm dot com>]

Hi Namhyung,

On 09/06/2013 12:07 PM, Namhyung Kim wrote:
> Hi Hemant,
>
> On Wed, 04 Sep 2013 23:07:57 +0530, Hemant wrote:
> > On 09/04/2013 12:12 PM, Namhyung Kim wrote:
> > > On Tue, 03 Sep 2013 13:06:55 +0530, Hemant Kumar wrote:
> > > > +	/*
> > > > +	 * Look for Section type = SHT_NOTE, flags = no SHF_ALLOC
> > > > +	 * and name = .note.stapsdt
> > > > +	 */
> > > > +	scn = elf_section_by_name(elf, &ehdr, &shdr, NOTE_SCN, NULL);
> > > > +	if (scn == NULL) {
> > > > +		pr_err("%s section not found!\n", NOTE_SCN);
> > > > +		goto out_end;
> > > > +	}
> > > > +
> > > > +	if (!(shdr.sh_type == SHT_NOTE) || (shdr.sh_flags & SHF_ALLOC))
> > > > +		goto out_end;
> > > > +
> > > > +	data = elf_getdata(scn, NULL);
> > > > +
> > > > +	/* Get the notes */
> > > > +	for (offset = 0; (next = gelf_getnote(data, offset, &nhdr, &name_off,
> > > > +					      &desc_off)) > 0; offset = next) {
> > > > +		tmp = populate_note(&elf, (const char *)((long)(data->d_buf) +
> > > > +							 (long)desc_off),
> > > > +				    nhdr.n_descsz, nhdr.n_type);
> > > Shouldn't we check the name of note being "stapsdt" as well as version
> > > (type) 3?
> > Since, we are already fetching the section NOTE_SCN (".note.stapsdt")
> > and then we check for the type being SHT_NOTE and SHF_ALLOC, is it
> > required to do the same for the individual notes?
> I don't know.  Now it seems only includes SDT notes with name being
> "stapsdt" and type being 3.  But things can be changed in future..
>
> Thanks,
> Namhyung

Yeah that may be a case. Ok, will put a check on the individual markers 
for their type.

Thanks
Hemant