Re: [PATCH 2/5] perf unwind: Do not look at globals

[Arnaldo Carvalho de Melo <acme@kernel.org>]

Em Tue, Jan 16, 2018 at 08:30:35PM +0100, Jiri Olsa escreveu:
> On Tue, Jan 16, 2018 at 12:36:21PM -0300, Arnaldo Carvalho de Melo wrote:
> > Em Tue, Jan 16, 2018 at 04:19:15PM +0100, Jiri Olsa escreveu:
> > > On Tue, Jan 16, 2018 at 11:24:35AM -0300, Arnaldo Carvalho de Melo wrote:
> > > 
> > > SNIP
> > > 
> > > > Cc: Adrian Hunter <adrian.hunter@intel.com>
> > > > Cc: David Ahern <dsahern@gmail.com>
> > > > Cc: Hendrick Brueckner <brueckner@linux.vnet.ibm.com>
> > > > Cc: Jiri Olsa <jolsa@kernel.org>
> > > > Cc: Namhyung Kim <namhyung@kernel.org>
> > > > Cc: Thomas Richter <tmricht@linux.vnet.ibm.com>
> > > > Cc: Wang Nan <wangnan0@huawei.com>
> > > > Link: https://lkml.kernel.org/n/tip-skbth8ufepbtw8xar7gdsb6l@git.kernel.org
> > > > Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
> > > > ---
> > > >  tools/perf/util/unwind-libunwind-local.c | 9 ---------
> > > >  1 file changed, 9 deletions(-)
> > > > 
> > > > diff --git a/tools/perf/util/unwind-libunwind-local.c b/tools/perf/util/unwind-libunwind-local.c
> > > > index 7a42f703e858..02dc5a9d8f72 100644
> > > > --- a/tools/perf/util/unwind-libunwind-local.c
> > > > +++ b/tools/perf/util/unwind-libunwind-local.c
> > > > @@ -631,9 +631,6 @@ static unw_accessors_t accessors = {
> > > >  
> > > >  static int _unwind__prepare_access(struct thread *thread)
> > > >  {
> > > > -	if (callchain_param.record_mode != CALLCHAIN_DWARF)
> > > > -		return 0;
> > > > -
> > > 
> > > this would create thread->addr_space also for data without
> > > dwarf callchains data, so I think we need to keep it
> > > 
> > > it should get set in apply_config_terms which calls parse_callchain_record
> > 
> > No, it should not set the global parameter, as this is just for a
> > specific event, i.e. we can have something like:
> > 
> > 	perf record -e cycles/call-graph=dwarf/ \
> > 		    -e instructions/call-graph=lbr/
> > 		    -e cache-misses/call-graph=fp
> > 
> > And then get these samples for the same thread.
> > 
> > If you want to avoid creating thread->addr_space, and we do want that,
> > sure, a followup patch should address that, we need to postpone
> > allocating it till we get a DWARF callchain in a sample for that
> > specific thread, when we then should allocate thread->addr_space.
> 
> yea that could be fixed.. however current code prevents
> to create that for data without DWARF data, and your change
> forces it

Humm? Which code?

[root@jouet ~]# perf record -e cycles/call-graph=dwarf/ -e instructions/call-graph=lbr/ -e cache-misses/call-graph=fp/ sleep 2
[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.140 MB perf.data (43 samples) ]
[root@jouet ~]# perf evlist -v
cycles/call-graph=dwarf/: size: 112, { sample_period, sample_freq }: 4000, sample_type: IP|TID|TIME|ADDR|CALLCHAIN|PERIOD|REGS_USER|STACK_USER|IDENTIFIER|DATA_SRC, read_format: ID, disabled: 1, inherit: 1, mmap: 1, comm: 1, freq: 1, enable_on_exec: 1, task: 1, mmap_data: 1, sample_id_all: 1, exclude_guest: 1, exclude_callchain_user: 1, mmap2: 1, comm_exec: 1, sample_regs_user: 0xff0fff, sample_stack_user: 8192
instructions/call-graph=lbr/: size: 112, config: 0x1, { sample_period, sample_freq }: 4000, sample_type: IP|TID|TIME|CALLCHAIN|PERIOD|BRANCH_STACK|IDENTIFIER, read_format: ID, disabled: 1, inherit: 1, freq: 1, enable_on_exec: 1, sample_id_all: 1, exclude_guest: 1, branch_sample_type: USER|CALL_STACK|NO_FLAGS|NO_CYCLES
cache-misses/call-graph=fp/: size: 112, config: 0x3, { sample_period, sample_freq }: 4000, sample_type: IP|TID|TIME|CALLCHAIN|PERIOD|IDENTIFIER, read_format: ID, disabled: 1, inherit: 1, freq: 1, enable_on_exec: 1, sample_id_all: 1, exclude_guest: 1
[root@jouet ~]# perf report --no-header | head -20
no symbols found in /usr/bin/sleep, maybe install a debug package?
# To display the perf.data header info, please use --header/--header-only options.
#
#
# Total Lost Samples: 0
#
# Samples: 15  of event 'cycles/call-graph=dwarf/'
# Event count (approx.): 2153016
#
# Children      Self  Command  Shared Object     Symbol                                 
# ........  ........  .......  ................  .......................................
#
    48.98%     0.00%  sleep    ld-2.26.so        [.] _start
            |
            ---_start
               |          
               |--35.15%--_dl_start
               |          _dl_sysdep_start
               |          dl_main
               |          |          
               |          |--18.44%--_dl_relocate_object
[root@jouet ~]#

- Arnaldo