| Message ID | 20230811081239.246365-1-nashuiliang@gmail.com (mailing list archive) |
|---|---|
| State | Superseded |
| Headers | show |
| Series | [v2] tracing/eprobe: Iterate trace_eprobe directly | expand |
On Fri, 11 Aug 2023 16:12:39 +0800 Chuang Wang <nashuiliang@gmail.com> wrote: > Refer to the description in [1], we can skip "container_of()" following > "list_for_each_entry()" by using "list_for_each_entry()" with > "struct trace_eprobe" and "tp.list". > > Also, this patch defines "for_each_trace_eprobe_on_trace_probe" to > simplify the code of the same logic. > > [1] https://lore.kernel.org/all/CAHk-=wjakjw6-rDzDDBsuMoDCqd+9ogifR_EE1F0K-jYek1CdA@mail.gmail.com/ > This looks good to me. Masami Hiramatsu (Google) <mhiramat@kernel.org> Thanks! > Signed-off-by: Chuang Wang <nashuiliang@gmail.com> > --- > v1 -> v2: > - add "for_each_trace_eprobe_on_trace_probe" as suggested by Masami > > v0 -> v1: > - remove "Fixes" in the commit information. > > kernel/trace/trace_eprobe.c | 18 +++++++++--------- > 1 file changed, 9 insertions(+), 9 deletions(-) > > diff --git a/kernel/trace/trace_eprobe.c b/kernel/trace/trace_eprobe.c > index a0a704ba27db..b7c6179f8b7f 100644 > --- a/kernel/trace/trace_eprobe.c > +++ b/kernel/trace/trace_eprobe.c > @@ -41,6 +41,10 @@ struct eprobe_data { > struct trace_eprobe *ep; > }; > > + > +#define for_each_trace_eprobe_on_trace_probe(ep, _tp) \ > + list_for_each_entry(ep, trace_probe_probe_list(_tp), tp.list) > + > static int __trace_eprobe_create(int argc, const char *argv[]); > > static void trace_event_probe_cleanup(struct trace_eprobe *ep) > @@ -640,7 +644,7 @@ static int disable_eprobe(struct trace_eprobe *ep, > static int enable_trace_eprobe(struct trace_event_call *call, > struct trace_event_file *file) > { > - struct trace_probe *pos, *tp; > + struct trace_probe *tp; > struct trace_eprobe *ep; > bool enabled; > int ret = 0; > @@ -662,8 +666,7 @@ static int enable_trace_eprobe(struct trace_event_call *call, > if (enabled) > return 0; > > - list_for_each_entry(pos, trace_probe_probe_list(tp), list) { > - ep = container_of(pos, struct trace_eprobe, tp); > + for_each_trace_eprobe_on_trace_probe(ep, tp) { > ret = enable_eprobe(ep, file); > if (ret) > break; > @@ -680,8 +683,7 @@ static int enable_trace_eprobe(struct trace_event_call *call, > */ > WARN_ON_ONCE(ret != -ENOMEM); > > - list_for_each_entry(pos, trace_probe_probe_list(tp), list) { > - ep = container_of(pos, struct trace_eprobe, tp); > + for_each_trace_eprobe_on_trace_probe(ep, tp) { > disable_eprobe(ep, file->tr); > if (!--cnt) > break; > @@ -699,7 +701,7 @@ static int enable_trace_eprobe(struct trace_event_call *call, > static int disable_trace_eprobe(struct trace_event_call *call, > struct trace_event_file *file) > { > - struct trace_probe *pos, *tp; > + struct trace_probe *tp; > struct trace_eprobe *ep; > > tp = trace_probe_primary_from_call(call); > @@ -716,10 +718,8 @@ static int disable_trace_eprobe(struct trace_event_call *call, > trace_probe_clear_flag(tp, TP_FLAG_PROFILE); > > if (!trace_probe_is_enabled(tp)) { > - list_for_each_entry(pos, trace_probe_probe_list(tp), list) { > - ep = container_of(pos, struct trace_eprobe, tp); > + for_each_trace_eprobe_on_trace_probe(ep, tp) > disable_eprobe(ep, file->tr); > - } > } > > out: > -- > 2.39.2 >
On Fri, 11 Aug 2023 16:12:39 +0800 Chuang Wang <nashuiliang@gmail.com> wrote: > + > +#define for_each_trace_eprobe_on_trace_probe(ep, _tp) \ > + list_for_each_entry(ep, trace_probe_probe_list(_tp), tp.list) > + As I replied to the other patch, but after this one was sent (sorry, I was hyperfocused on my work the last couple of days and was not reading email, so my INBOX is a bit full). I really hate the verbosity of that name. At a minimum, let's call it: for_each_trace_point_eprobe() -- Steve
On Fri, 11 Aug 2023 15:45:23 -0400 Steven Rostedt <rostedt@goodmis.org> wrote: > On Fri, 11 Aug 2023 16:12:39 +0800 > Chuang Wang <nashuiliang@gmail.com> wrote: > > > + > > +#define for_each_trace_eprobe_on_trace_probe(ep, _tp) \ > > + list_for_each_entry(ep, trace_probe_probe_list(_tp), tp.list) > > + > > As I replied to the other patch, but after this one was sent (sorry, I was > hyperfocused on my work the last couple of days and was not reading email, > so my INBOX is a bit full). I really hate the verbosity of that name. > > At a minimum, let's call it: > > for_each_trace_point_eprobe() OK, what about "for_each_trace_eprobe_on()"? I would like to clarify - what type is returned - not all trace_eprobes, but only on the trace_probe. Thank you, > > -- Steve
On Sat, 12 Aug 2023 14:20:19 +0900 Masami Hiramatsu (Google) <mhiramat@kernel.org> wrote: > > At a minimum, let's call it: > > > > for_each_trace_point_eprobe() > > OK, what about "for_each_trace_eprobe_on()"? I would like to clarify > > - what type is returned > - not all trace_eprobes, but only on the trace_probe. > > Thank you, for_each_trace_eprobe_tp() or for_each_trace_tp_eprobe() ? As it only works for a trace_probe. -- Steve
On Wed, 16 Aug 2023 16:08:50 -0400 Steven Rostedt <rostedt@goodmis.org> wrote: > On Sat, 12 Aug 2023 14:20:19 +0900 > Masami Hiramatsu (Google) <mhiramat@kernel.org> wrote: > > > > At a minimum, let's call it: > > > > > > for_each_trace_point_eprobe() > > > > OK, what about "for_each_trace_eprobe_on()"? I would like to clarify > > > > - what type is returned > > - not all trace_eprobes, but only on the trace_probe. > > > > Thank you, > > for_each_trace_eprobe_tp() or for_each_trace_tp_eprobe() ? for_each_trace_eprobe_tp() is OK for me. Thanks! > > As it only works for a trace_probe. > > -- Steve
OK, I will submit a new patch using "for_each_trace_eprobe_tp". On Fri, Aug 18, 2023 at 7:03 PM Masami Hiramatsu <mhiramat@kernel.org> wrote: > > On Wed, 16 Aug 2023 16:08:50 -0400 > Steven Rostedt <rostedt@goodmis.org> wrote: > > > On Sat, 12 Aug 2023 14:20:19 +0900 > > Masami Hiramatsu (Google) <mhiramat@kernel.org> wrote: > > > > > > At a minimum, let's call it: > > > > > > > > for_each_trace_point_eprobe() > > > > > > OK, what about "for_each_trace_eprobe_on()"? I would like to clarify > > > > > > - what type is returned > > > - not all trace_eprobes, but only on the trace_probe. > > > > > > Thank you, > > > > for_each_trace_eprobe_tp() or for_each_trace_tp_eprobe() ? > > for_each_trace_eprobe_tp() is OK for me. > > Thanks! > > > > > As it only works for a trace_probe. > > > > -- Steve > > > -- > Masami Hiramatsu (Google) <mhiramat@kernel.org>
diff --git a/kernel/trace/trace_eprobe.c b/kernel/trace/trace_eprobe.c index a0a704ba27db..b7c6179f8b7f 100644 --- a/kernel/trace/trace_eprobe.c +++ b/kernel/trace/trace_eprobe.c @@ -41,6 +41,10 @@ struct eprobe_data { struct trace_eprobe *ep; }; + +#define for_each_trace_eprobe_on_trace_probe(ep, _tp) \ + list_for_each_entry(ep, trace_probe_probe_list(_tp), tp.list) + static int __trace_eprobe_create(int argc, const char *argv[]); static void trace_event_probe_cleanup(struct trace_eprobe *ep) @@ -640,7 +644,7 @@ static int disable_eprobe(struct trace_eprobe *ep, static int enable_trace_eprobe(struct trace_event_call *call, struct trace_event_file *file) { - struct trace_probe *pos, *tp; + struct trace_probe *tp; struct trace_eprobe *ep; bool enabled; int ret = 0; @@ -662,8 +666,7 @@ static int enable_trace_eprobe(struct trace_event_call *call, if (enabled) return 0; - list_for_each_entry(pos, trace_probe_probe_list(tp), list) { - ep = container_of(pos, struct trace_eprobe, tp); + for_each_trace_eprobe_on_trace_probe(ep, tp) { ret = enable_eprobe(ep, file); if (ret) break; @@ -680,8 +683,7 @@ static int enable_trace_eprobe(struct trace_event_call *call, */ WARN_ON_ONCE(ret != -ENOMEM); - list_for_each_entry(pos, trace_probe_probe_list(tp), list) { - ep = container_of(pos, struct trace_eprobe, tp); + for_each_trace_eprobe_on_trace_probe(ep, tp) { disable_eprobe(ep, file->tr); if (!--cnt) break; @@ -699,7 +701,7 @@ static int enable_trace_eprobe(struct trace_event_call *call, static int disable_trace_eprobe(struct trace_event_call *call, struct trace_event_file *file) { - struct trace_probe *pos, *tp; + struct trace_probe *tp; struct trace_eprobe *ep; tp = trace_probe_primary_from_call(call); @@ -716,10 +718,8 @@ static int disable_trace_eprobe(struct trace_event_call *call, trace_probe_clear_flag(tp, TP_FLAG_PROFILE); if (!trace_probe_is_enabled(tp)) { - list_for_each_entry(pos, trace_probe_probe_list(tp), list) { - ep = container_of(pos, struct trace_eprobe, tp); + for_each_trace_eprobe_on_trace_probe(ep, tp) disable_eprobe(ep, file->tr); - } } out:
Refer to the description in [1], we can skip "container_of()" following "list_for_each_entry()" by using "list_for_each_entry()" with "struct trace_eprobe" and "tp.list". Also, this patch defines "for_each_trace_eprobe_on_trace_probe" to simplify the code of the same logic. [1] https://lore.kernel.org/all/CAHk-=wjakjw6-rDzDDBsuMoDCqd+9ogifR_EE1F0K-jYek1CdA@mail.gmail.com/ Signed-off-by: Chuang Wang <nashuiliang@gmail.com> --- v1 -> v2: - add "for_each_trace_eprobe_on_trace_probe" as suggested by Masami v0 -> v1: - remove "Fixes" in the commit information. kernel/trace/trace_eprobe.c | 18 +++++++++--------- 1 file changed, 9 insertions(+), 9 deletions(-)