On Mon, Feb 19, 2018 at 11:20:36AM +0900, Namhyung Kim wrote:
SNIP
> > +static void machine__set_kernel_mmap(struct machine *machine,
> > + u64 start, u64 end)
> > +{
> > + int i;
> > +
> > + for (i = 0; i < MAP__NR_TYPES; i++) {
> > + machine->vmlinux_maps[i]->start = start;
> > + machine->vmlinux_maps[i]->end = end;
> > +
> > + /*
> > + * Be a bit paranoid here, some perf.data file came with
> > + * a zero sized synthesized MMAP event for the kernel.
> > + */
> > + if (machine->vmlinux_maps[i]->end == 0)
> > + machine->vmlinux_maps[i]->end = ~0ULL;
>
> Hmm.. this will make map_groups__fixup_end() below not working since
> it only updates if the end address of a map is zero.
>
> And about the paranoid check, AFAIK the only case it cares is the
> machine__process_kernel_mmap_event() which calls it with
> event->mmap.start and event->mmap.start + event->mmap.len. Thus, in
> order for the end to be zero, both start and len should be zero. Then
> I think this condition can be changed like below:
>
> diff --git a/tools/perf/util/machine.c b/tools/perf/util/machine.c
> index fe27ef55cbb9..c8acb603c359 100644
> --- a/tools/perf/util/machine.c
> +++ b/tools/perf/util/machine.c
> @@ -1226,7 +1226,7 @@ static void machine__set_kernel_mmap(struct machine
> *machine,
> * Be a bit paranoid here, some perf.data file came with
> * a zero sized synthesized MMAP event for the kernel.
> */
> - if (machine->vmlinux_maps[i]->end == 0)
> + if (start == 0 && end == 0)
> machine->vmlinux_maps[i]->end = ~0ULL;
> }
right, the call from machine__create_kernel_maps will have always
the start != 0 and there will be subsequent map_groups__fixup_end
call..
seems ok to me.. will you send the patch?
thanks,
jirka
