Re: [PATCH 1/8] perf tools: Fix bug in 'perf list event_glob'.

[Dongsheng Yang]

On 12/27/2013 02:21 PM, Arnaldo Carvalho de Melo wrote:
> Em Fri, Dec 27, 2013 at 06:03:27PM -0500, Dongsheng Yang escreveu:
> > As what the usage showing, we can input a event_glob as an argument to list
> > all symbols. But it does not work well right now.
> >
> > Example:
> > 	# perf list kvmmmu
> >
> > 	List of pre-defined events (to be used in -e):
> >
> > This patch make it work well as follows.
> It works:
>
> List of pre-defined events (to be used in -e):
> [root@zoo ~]# perf list kvmmmu*:*
>    kvmmmu:kvm_mmu_pagetable_walk                      [Tracepoint event]
>    kvmmmu:kvm_mmu_paging_element                      [Tracepoint event]
>    kvmmmu:kvm_mmu_set_accessed_bit                    [Tracepoint event]
>    kvmmmu:kvm_mmu_set_dirty_bit                       [Tracepoint event]
>    kvmmmu:kvm_mmu_walker_error                        [Tracepoint event]
>    kvmmmu:kvm_mmu_get_page                            [Tracepoint event]
>    kvmmmu:kvm_mmu_sync_page                           [Tracepoint event]
>    kvmmmu:kvm_mmu_unsync_page                         [Tracepoint event]
>    kvmmmu:kvm_mmu_prepare_zap_page                    [Tracepoint event]
>    kvmmmu:mark_mmio_spte                              [Tracepoint event]
>    kvmmmu:handle_mmio_page_fault                      [Tracepoint event]
>    kvmmmu:fast_page_fault                             [Tracepoint event]
>    kvmmmu:kvm_mmu_invalidate_zap_all_pages            [Tracepoint event]
>    kvmmmu:check_mmio_spte                             [Tracepoint event]
> [root@zoo ~]#
>
> The documentation says, as you repeated above, that a glob can be
> provided, not a _substring_ :-)

Yes, you are right. Sorry for that I misunderstood the word of 'glob'.

Thank you for your explanation. :)

Yang
> To achive a result similar to:
>
> [root@zoo ~]# perf list hw
>    cpu-cycles OR cycles                               [Hardware event]
>    instructions                                       [Hardware event]
>    cache-references                                   [Hardware event]
>    cache-misses                                       [Hardware event]
>    branch-instructions OR branches                    [Hardware event]
>    branch-misses                                      [Hardware event]
>    bus-cycles                                         [Hardware event]
>    stalled-cycles-frontend OR idle-cycles-frontend    [Hardware event]
>    ref-cycles                                         [Hardware event]
> [root@zoo ~]#
>
> We could say that if no ':' is found in the provided string, and it
> doesn't matches one of the other classes, i.e. hw, sw, cache, etc, and
> it is the name of a tracepoint subsystem, then it would be equivalent
> to 'tracepoint_subsys:*', i.e. 'perf list skb' woud be equivalent to:
>
> [root@zoo ~]# perf list skb:*
>    skb:kfree_skb                                      [Tracepoint event]
>    skb:consume_skb                                    [Tracepoint event]
>    skb:skb_copy_datagram_iovec                        [Tracepoint event]
> [root@zoo ~]#
>
> - Arnaldo
>   
> > Verification:
> > 	# perf list kvmmmu
> >
> > 	List of pre-defined events (to be used in -e):
> > 	  kvmmmu:kvm_mmu_pagetable_walk                      [Tracepoint event]
> > 	  kvmmmu:kvm_mmu_paging_element                      [Tracepoint event]
> > 	  kvmmmu:kvm_mmu_set_accessed_bit                    [Tracepoint event]
> > 	  kvmmmu:kvm_mmu_set_dirty_bit                       [Tracepoint event]
> > 	  kvmmmu:kvm_mmu_walker_error                        [Tracepoint event]
> > 	  kvmmmu:kvm_mmu_get_page                            [Tracepoint event]
> > 	  kvmmmu:kvm_mmu_sync_page                           [Tracepoint event]
> > 	  kvmmmu:kvm_mmu_unsync_page                         [Tracepoint event]
> > 	  kvmmmu:kvm_mmu_prepare_zap_page                    [Tracepoint event]
> > 	  kvmmmu:mark_mmio_spte                              [Tracepoint event]
> > 	  kvmmmu:handle_mmio_page_fault                      [Tracepoint event]
> > 	  kvmmmu:fast_page_fault                             [Tracepoint event]
> >
> > Signed-off-by: Dongsheng Yang <yangds.fnst@xxxxxxxxxxxxxx>
> > ---
> >   tools/perf/util/parse-events.c | 4 ++--
> >   1 file changed, 2 insertions(+), 2 deletions(-)
> >
> > diff --git a/tools/perf/util/parse-events.c b/tools/perf/util/parse-events.c
> > index 094c28b..d887ea4 100644
> > --- a/tools/perf/util/parse-events.c
> > +++ b/tools/perf/util/parse-events.c
> > @@ -1236,6 +1236,8 @@ void print_events(const char *event_glob, bool name_only)
> >   
> >   	print_pmu_events(event_glob, name_only);
> >   
> > +	print_tracepoint_events(event_glob, NULL, name_only);
> > +
> >   	if (event_glob != NULL)
> >   		return;
> >   
> > @@ -1254,8 +1256,6 @@ void print_events(const char *event_glob, bool name_only)
> >   			event_type_descriptors[PERF_TYPE_BREAKPOINT]);
> >   		printf("\n");
> >   	}
> > -
> > -	print_tracepoint_events(NULL, NULL, name_only);
> >   }
> >   
> >   int parse_events__is_hardcoded_term(struct parse_events_term *term)
> > --
> > 1.8.2.1
> --
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@xxxxxxxxxxxxxxx
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/