Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.kernel > #1253502 > unrolled thread

[PATCH 1/3] perf tools: Move callchain help messages to callchain.h

Started byNamhyung Kim <namhyung@kernel.org>
First post2015-10-22 08:30 +0200
Last post2015-10-23 10:40 +0200
Articles 9 — 4 participants

Back to article view | Back to linux.kernel

Contents

  [PATCH 1/3] perf tools: Move callchain help messages to callchain.h Namhyung Kim <namhyung@kernel.org> - 2015-10-22 08:30 +0200
    Re: [PATCH 1/3] perf tools: Move callchain help messages to  callchain.h Ingo Molnar <mingo@kernel.org> - 2015-10-22 10:10 +0200
      Re: [PATCH 1/3] perf tools: Move callchain help messages to callchain.h Namhyung Kim <namhyung@kernel.org> - 2015-10-22 14:20 +0200
        Re: [PATCH 1/3] perf tools: Move callchain help messages to  callchain.h Arnaldo Carvalho de Melo <acme@kernel.org> - 2015-10-22 16:20 +0200
        Re: [PATCH 1/3] perf tools: Move callchain help messages to  callchain.h Ingo Molnar <mingo@kernel.org> - 2015-10-23 12:00 +0200
          Re: [PATCH 1/3] perf tools: Move callchain help messages to  callchain.h Arnaldo Carvalho de Melo <acme@kernel.org> - 2015-10-23 16:30 +0200
            Re: [PATCH 1/3] perf tools: Move callchain help messages to  callchain.h Arnaldo Carvalho de Melo <acme@kernel.org> - 2015-10-23 18:50 +0200
    Re: [PATCH 1/3] perf tools: Move callchain help messages to  callchain.h Arnaldo Carvalho de Melo <acme@kernel.org> - 2015-10-22 20:40 +0200
    [tip:perf/core] perf tools:   Move callchain help messages to callchain.h tip-bot for Namhyung Kim <tipbot@zytor.com> - 2015-10-23 10:40 +0200

#1253502 — [PATCH 1/3] perf tools: Move callchain help messages to callchain.h

FromNamhyung Kim <namhyung@kernel.org>
Date2015-10-22 08:30 +0200
Subject[PATCH 1/3] perf tools: Move callchain help messages to callchain.h
Message-ID<qmhxU-2nM-7@gated-at.bofh.it>
These messages will be used by 'perf top' in the next patch.

Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Borislav Petkov <bp@suse.de>
Cc: Chandler Carruth <chandlerc@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Stephane Eranian <eranian@google.com>
Cc: Wang Nan <wangnan0@huawei.com>
Cc: Brendan Gregg <brendan.d.gregg@gmail.com>
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
---
 tools/perf/builtin-record.c |  8 +-------
 tools/perf/builtin-report.c | 10 +++++++---
 tools/perf/util/callchain.h | 12 ++++++++++++
 3 files changed, 20 insertions(+), 10 deletions(-)

diff --git a/tools/perf/builtin-record.c b/tools/perf/builtin-record.c
index 24ace2f318c1..1a117623d396 100644
--- a/tools/perf/builtin-record.c
+++ b/tools/perf/builtin-record.c
@@ -1010,13 +1010,7 @@ static struct record record = {
 	},
 };
 
-#define CALLCHAIN_HELP "setup and enables call-graph (stack chain/backtrace) recording: "
-
-#ifdef HAVE_DWARF_UNWIND_SUPPORT
-const char record_callchain_help[] = CALLCHAIN_HELP "fp dwarf lbr";
-#else
-const char record_callchain_help[] = CALLCHAIN_HELP "fp lbr";
-#endif
+const char record_callchain_help[] = CALLCHAIN_RECORD_HELP;
 
 /*
  * XXX Will stay a global variable till we fix builtin-script.c to stop messing
diff --git a/tools/perf/builtin-report.c b/tools/perf/builtin-report.c
index 3b23b25d1589..18a8c52d921e 100644
--- a/tools/perf/builtin-report.c
+++ b/tools/perf/builtin-report.c
@@ -625,6 +625,9 @@ parse_percent_limit(const struct option *opt, const char *str,
 	return 0;
 }
 
+const char report_callchain_help[] = "Display callchains using " CALLCHAIN_REPORT_HELP ". "
+				     "Default: graph,0.5,caller";
+
 int cmd_report(int argc, const char **argv, const char *prefix __maybe_unused)
 {
 	struct perf_session *session;
@@ -699,9 +702,10 @@ int cmd_report(int argc, const char **argv, const char *prefix __maybe_unused)
 		   "regex filter to identify parent, see: '--sort parent'"),
 	OPT_BOOLEAN('x', "exclude-other", &symbol_conf.exclude_other,
 		    "Only display entries with parent-match"),
-	OPT_CALLBACK_DEFAULT('g', "call-graph", &report, "output_type,min_percent[,print_limit],call_order[,branch]",
-		     "Display callchains using output_type (graph, flat, fractal, or none) , min percent threshold, optional print limit, callchain order, key (function or address), add branches. "
-		     "Default: graph,0.5,caller", &report_parse_callchain_opt, callchain_default_opt),
+	OPT_CALLBACK_DEFAULT('g', "call-graph", &report,
+			     "output_type,min_percent[,print_limit],call_order[,branch]",
+			     report_callchain_help, &report_parse_callchain_opt,
+			     callchain_default_opt),
 	OPT_BOOLEAN(0, "children", &symbol_conf.cumulate_callchain,
 		    "Accumulate callchains of children and show total overhead as well"),
 	OPT_INTEGER(0, "max-stack", &report.max_stack,
diff --git a/tools/perf/util/callchain.h b/tools/perf/util/callchain.h
index acee2b3cd801..c9e3a2e85a72 100644
--- a/tools/perf/util/callchain.h
+++ b/tools/perf/util/callchain.h
@@ -7,6 +7,18 @@
 #include "event.h"
 #include "symbol.h"
 
+#define CALLCHAIN_HELP "setup and enables call-graph (stack chain/backtrace) recording: "
+
+#ifdef HAVE_DWARF_UNWIND_SUPPORT
+#define CALLCHAIN_RECORD_HELP  CALLCHAIN_HELP "fp dwarf lbr"
+#else
+#define CALLCHAIN_RECORD_HELP  CALLCHAIN_HELP "fp lbr"
+#endif
+
+#define CALLCHAIN_REPORT_HELP  "output_type (graph, flat, fractal, or none), " \
+	"min percent threshold, optional print limit, callchain order, " \
+	"key (function or address), add branches"
+
 enum perf_call_graph_mode {
 	CALLCHAIN_NONE,
 	CALLCHAIN_FP,
-- 
2.6.0

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

[toc] | [next] | [standalone]

#1253562 — Re: [PATCH 1/3] perf tools: Move callchain help messages to callchain.h

FromIngo Molnar <mingo@kernel.org>
Date2015-10-22 10:10 +0200
SubjectRe: [PATCH 1/3] perf tools: Move callchain help messages to callchain.h
Message-ID<qmj6F-4MD-3@gated-at.bofh.it>
In reply to#1253502
* Namhyung Kim <namhyung@kernel.org> wrote:

> +#define CALLCHAIN_HELP "setup and enables call-graph (stack chain/backtrace) recording: "
> +
> +#ifdef HAVE_DWARF_UNWIND_SUPPORT
> +#define CALLCHAIN_RECORD_HELP  CALLCHAIN_HELP "fp dwarf lbr"
> +#else
> +#define CALLCHAIN_RECORD_HELP  CALLCHAIN_HELP "fp lbr"
> +#endif

nano-nit, could we structure such balanced #ifdefs the following way:

#ifdef HAVE_DWARF_UNWIND_SUPPORT
# define CALLCHAIN_RECORD_HELP CALLCHAIN_HELP "fp dwarf lbr"
#else
# define CALLCHAIN_RECORD_HELP CALLCHAIN_HELP "fp lbr"
#endif

makes the construct stand out a lot better visually.

I also had another look at the help text:

> output_type,min_percent[,print_limit],call_order[,branch]

> +#define CALLCHAIN_REPORT_HELP  "output_type (graph, flat, fractal, or none), " \
> +	"min percent threshold, optional print limit, callchain order, " \
> +	"key (function or address), add branches"

Btw., when I first read this message in the help text yesterday, I had to read the 
'min percent threshold' twice, to realize that the default 0.5 is in units of 
percentage - the wording wasn't entirely clear about that.

Also, I had to go into the code to decode the real meaning of all the other 
parameters. I'd have expected them to be more obvious from reading the help text.

Wording them the following way would have made things a lot more apparent to me:

	print_style,min_percent[,print_percent],call_order[,key]

	call chain tree printing style (graph|flat|fractal|none)
	minimum tree inclusion threshold (percent)
	printing threshold (percent)
	call chain order (caller|callee)
	key (function|address|branch)

Note that I extended the help text with new options not mentioned in the help text 
but present in the current code - such as the 'branch' key.

Also note that in the code I did not find any trace of the '[,branch]' and
'add branches' part present in the help text. What we have is a 'branch'
option in the (optional) key parameter.

I also made various edits to the help text to make it more consistent and more 
self-explanatory. I think we should also put the various options into a new line 
in the help screen, not the single line dump of text it is currently.

Btw., we also have a grammar problem with all things call chains: there's 800+ 
occurances of 'callchain' in the perf code, and less than 20 spellings of 'call 
chain'. But the latter is the correct variant: Google won't even let you search 
for 'callchain' by default and corrects it to 'call chain' automatically.

If you insist on searching for 'callchain', Google finds this number of hits:

  'code callchain':       54,200
  'code call chain': 141,000,000

I think it's pretty obvious what the dominant spelling is in the industry! ;-)

So we should probably rename all occurances of 'callchain' to 'call chain' or 
'call_chain'.

Thanks,

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

[toc] | [prev] | [next] | [standalone]

#1253748

FromNamhyung Kim <namhyung@kernel.org>
Date2015-10-22 14:20 +0200
Message-ID<qmn0B-28d-1@gated-at.bofh.it>
In reply to#1253562
On Thu, Oct 22, 2015 at 5:02 PM, Ingo Molnar <mingo@kernel.org> wrote:
>
> * Namhyung Kim <namhyung@kernel.org> wrote:
>
>> +#define CALLCHAIN_HELP "setup and enables call-graph (stack chain/backtrace) recording: "
>> +
>> +#ifdef HAVE_DWARF_UNWIND_SUPPORT
>> +#define CALLCHAIN_RECORD_HELP  CALLCHAIN_HELP "fp dwarf lbr"
>> +#else
>> +#define CALLCHAIN_RECORD_HELP  CALLCHAIN_HELP "fp lbr"
>> +#endif
>
> nano-nit, could we structure such balanced #ifdefs the following way:
>
> #ifdef HAVE_DWARF_UNWIND_SUPPORT
> # define CALLCHAIN_RECORD_HELP CALLCHAIN_HELP "fp dwarf lbr"
> #else
> # define CALLCHAIN_RECORD_HELP CALLCHAIN_HELP "fp lbr"
> #endif
>
> makes the construct stand out a lot better visually.

OK

>
> I also had another look at the help text:
>
>> output_type,min_percent[,print_limit],call_order[,branch]
>
>> +#define CALLCHAIN_REPORT_HELP  "output_type (graph, flat, fractal, or none), " \
>> +     "min percent threshold, optional print limit, callchain order, " \
>> +     "key (function or address), add branches"
>
> Btw., when I first read this message in the help text yesterday, I had to read the
> 'min percent threshold' twice, to realize that the default 0.5 is in units of
> percentage - the wording wasn't entirely clear about that.

OK

>
> Also, I had to go into the code to decode the real meaning of all the other
> parameters. I'd have expected them to be more obvious from reading the help text.

Did you check the man page also?  I think we have (short) explanation
for each parameter and users should read it first to understand the
meaning.  But I agree that the help text should also be improved to
provide quick reference.


>
> Wording them the following way would have made things a lot more apparent to me:
>
>         print_style,min_percent[,print_percent],call_order[,key]
>
>         call chain tree printing style (graph|flat|fractal|none)
>         minimum tree inclusion threshold (percent)
>         printing threshold (percent)

Note that this 'printing threshold' is not percent.  It's to limit
number of callchain entries printed for each hist entry.  However it
works for --stdio only probably since it lacks interactive
collapse/expand feature.


>         call chain order (caller|callee)
>         key (function|address|branch)
>
> Note that I extended the help text with new options not mentioned in the help text
> but present in the current code - such as the 'branch' key.
>
> Also note that in the code I did not find any trace of the '[,branch]' and
> 'add branches' part present in the help text. What we have is a 'branch'
> option in the (optional) key parameter.

Looking at the document, it seems branch is not a key:

    branch can be:
    - branch: include last branch information in callgraph
    when available. Usually more convenient to use --branch-history
    for this.

Confusingly, it was checked in parse_callchain_sort_key() but does
nothing with the sort key IIUC.

>
> I also made various edits to the help text to make it more consistent and more
> self-explanatory. I think we should also put the various options into a new line
> in the help screen, not the single line dump of text it is currently.

OK

>
> Btw., we also have a grammar problem with all things call chains: there's 800+
> occurances of 'callchain' in the perf code, and less than 20 spellings of 'call
> chain'. But the latter is the correct variant: Google won't even let you search
> for 'callchain' by default and corrects it to 'call chain' automatically.
>
> If you insist on searching for 'callchain', Google finds this number of hits:
>
>   'code callchain':       54,200
>   'code call chain': 141,000,000
>
> I think it's pretty obvious what the dominant spelling is in the industry! ;-)
>
> So we should probably rename all occurances of 'callchain' to 'call chain' or
> 'call_chain'.

Not sure about this part.  Do you really think it's worth changing?

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

[toc] | [prev] | [next] | [standalone]

#1253860 — Re: [PATCH 1/3] perf tools: Move callchain help messages to callchain.h

FromArnaldo Carvalho de Melo <acme@kernel.org>
Date2015-10-22 16:20 +0200
SubjectRe: [PATCH 1/3] perf tools: Move callchain help messages to callchain.h
Message-ID<qmoSL-4U2-29@gated-at.bofh.it>
In reply to#1253748
Em Thu, Oct 22, 2015 at 09:13:49PM +0900, Namhyung Kim escreveu:
> On Thu, Oct 22, 2015 at 5:02 PM, Ingo Molnar <mingo@kernel.org> wrote:
> >
> > * Namhyung Kim <namhyung@kernel.org> wrote:
> >
> >> +#define CALLCHAIN_HELP "setup and enables call-graph (stack chain/backtrace) recording: "
> >> +
> >> +#ifdef HAVE_DWARF_UNWIND_SUPPORT
> >> +#define CALLCHAIN_RECORD_HELP  CALLCHAIN_HELP "fp dwarf lbr"
> >> +#else
> >> +#define CALLCHAIN_RECORD_HELP  CALLCHAIN_HELP "fp lbr"
> >> +#endif
> >
> > nano-nit, could we structure such balanced #ifdefs the following way:
> >
> > #ifdef HAVE_DWARF_UNWIND_SUPPORT
> > # define CALLCHAIN_RECORD_HELP CALLCHAIN_HELP "fp dwarf lbr"
> > #else
> > # define CALLCHAIN_RECORD_HELP CALLCHAIN_HELP "fp lbr"
> > #endif
> >
> > makes the construct stand out a lot better visually.
> 
> OK

Done.
 
> >
> > I also had another look at the help text:
> >
> >> output_type,min_percent[,print_limit],call_order[,branch]
> >
> >> +#define CALLCHAIN_REPORT_HELP  "output_type (graph, flat, fractal, or none), " \
> >> +     "min percent threshold, optional print limit, callchain order, " \
> >> +     "key (function or address), add branches"
> >
> > Btw., when I first read this message in the help text yesterday, I had to read the
> > 'min percent threshold' twice, to realize that the default 0.5 is in units of
> > percentage - the wording wasn't entirely clear about that.
> 
> OK
> 
> >
> > Also, I had to go into the code to decode the real meaning of all the other
> > parameters. I'd have expected them to be more obvious from reading the help text.
> 
> Did you check the man page also?  I think we have (short) explanation
> for each parameter and users should read it first to understand the
> meaning.  But I agree that the help text should also be improved to
> provide quick reference.
> 
> 
> >
> > Wording them the following way would have made things a lot more apparent to me:
> >
> >         print_style,min_percent[,print_percent],call_order[,key]
> >
> >         call chain tree printing style (graph|flat|fractal|none)
> >         minimum tree inclusion threshold (percent)
> >         printing threshold (percent)
> 
> Note that this 'printing threshold' is not percent.  It's to limit
> number of callchain entries printed for each hist entry.  However it
> works for --stdio only probably since it lacks interactive
> collapse/expand feature.
> 
> 
> >         call chain order (caller|callee)
> >         key (function|address|branch)
> >
> > Note that I extended the help text with new options not mentioned in the help text
> > but present in the current code - such as the 'branch' key.
> >
> > Also note that in the code I did not find any trace of the '[,branch]' and
> > 'add branches' part present in the help text. What we have is a 'branch'
> > option in the (optional) key parameter.
> 
> Looking at the document, it seems branch is not a key:
> 
>     branch can be:
>     - branch: include last branch information in callgraph
>     when available. Usually more convenient to use --branch-history
>     for this.
> 
> Confusingly, it was checked in parse_callchain_sort_key() but does
> nothing with the sort key IIUC.
> 
> >
> > I also made various edits to the help text to make it more consistent and more
> > self-explanatory. I think we should also put the various options into a new line
> > in the help screen, not the single line dump of text it is currently.
> 
> OK

These can come on a followup patch, improving the situation, right?
 
> >
> > Btw., we also have a grammar problem with all things call chains: there's 800+
> > occurances of 'callchain' in the perf code, and less than 20 spellings of 'call
> > chain'. But the latter is the correct variant: Google won't even let you search
> > for 'callchain' by default and corrects it to 'call chain' automatically.
> >
> > If you insist on searching for 'callchain', Google finds this number of hits:
> >
> >   'code callchain':       54,200
> >   'code call chain': 141,000,000
> >
> > I think it's pretty obvious what the dominant spelling is in the industry! ;-)
> >
> > So we should probably rename all occurances of 'callchain' to 'call chain' or
> > 'call_chain'.
> 
> Not sure about this part.  Do you really think it's worth changing?
> 
> Thanks,
> Namhyung
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

[toc] | [prev] | [next] | [standalone]

#1254444 — Re: [PATCH 1/3] perf tools: Move callchain help messages to callchain.h

FromIngo Molnar <mingo@kernel.org>
Date2015-10-23 12:00 +0200
SubjectRe: [PATCH 1/3] perf tools: Move callchain help messages to callchain.h
Message-ID<qmHiH-6aF-29@gated-at.bofh.it>
In reply to#1253748
* Namhyung Kim <namhyung@kernel.org> wrote:

> > Also, I had to go into the code to decode the real meaning of all the other 
> > parameters. I'd have expected them to be more obvious from reading the help 
> > text.
> 
> Did you check the man page also?  I think we have (short) explanation for each 
> parameter and users should read it first to understand the meaning.  But I agree 
> that the help text should also be improved to provide quick reference.

ah, no.

Is there a way to call up the manpage output for any given option, or so?

Such as:

   perf report -g help

Outputs the short-style help text you fixed:

 Usage: perf report [<options>]

    -g, --call-graph <print_type,threshold[,print_limit],order,sort_key[,branch]>
                          Display call graph (stack chain/backtrace):

                                print_type:     call graph printing style (graph|flat|fractal|none)
                                threshold:      minimum call graph inclusion threshold (<percent>)
                                print_limit:    maximum number of call graph entry (<number>)
                                order:          call graph order (caller|callee)
                                sort_key:       call graph sort key (function|address)
                                branch:         include last branch info to call graph (branch)

                                Default: graph,0.5,caller,function

and we could append one more line:

 # Suggestion: By typing 'perf report -g man' you can see the detailed manpage of option

OTOH the manpage is not easily parsed into per option sections, right? So we could 
bring up the whole manpage.

Btw., another usability detail I noticed yesterday is that when I typed 'perf 
report -h' I got so much output that I couldn't find the specific option I was 
looking for, because there's no apparent ordering of the output:

triton:~/tip> perf report -h 2>&1 | grep -e ' -[a-Z],'
    -i, --input <file>    input file name
    -v, --verbose         be more verbose (show symbol address, etc)
    -D, --dump-raw-trace  dump raw trace in ASCII
    -k, --vmlinux <file>  vmlinux pathname
    -f, --force           don't complain, do it
    -m, --modules         load module symbols - WARNING: use only with -k and LIVE kernel
    -n, --show-nr-samples
    -T, --threads         Show per-thread event counters
    -s, --sort <key[,key2...]>
    -F, --fields <key[,keys...]>
    -p, --parent <regex>  regex filter to identify parent, see: '--sort parent'
    -x, --exclude-other   Only display entries with parent-match
    -g, --call-graph <print_type,threshold[,print_limit],order,sort_key[,branch]>
    -G, --inverted        alias for inverted call graph
    -d, --dsos <dso[,dso...]>
    -c, --comms <comm[,comm...]>
    -S, --symbols <symbol[,symbol...]>
    -w, --column-widths <width[,width...]>
    -t, --field-separator <separator>
    -U, --hide-unresolved
    -C, --cpu <cpu>       list of cpus to profile
    -I, --show-info       Display extended information about perf.data file
    -M, --disassembler-style <disassembler style>
    -b, --branch-stack    use branch records for per branch histogram filling

Such (alphabetic) ordering would be easier to navigate:

    -b, --branch-stack    use branch records for per branch histogram filling
    -c, --comms <comm[,comm...]>
    -C, --cpu <cpu>       list of cpus to profile
    -d, --dsos <dso[,dso...]>
    -D, --dump-raw-trace  dump raw trace in ASCII
    -F, --fields <key[,keys...]>
    -f, --force           don't complain, do it
    -g, --call-graph <print_type,threshold[,print_limit],order,sort_key[,branch]>
    -G, --inverted        alias for inverted call graph
    -i, --input <file>    input file name
    -I, --show-info       Display extended information about perf.data file
    -k, --vmlinux <file>  vmlinux pathname
    -M, --disassembler-style <disassembler style>
    -m, --modules         load module symbols - WARNING: use only with -k and LIVE kernel
    -n, --show-nr-samples
    -p, --parent <regex>  regex filter to identify parent, see: '--sort parent'
    -s, --sort <key[,key2...]>
    -S, --symbols <symbol[,symbol...]>
    -t, --field-separator <separator>
    -T, --threads         Show per-thread event counters
    -U, --hide-unresolved
    -v, --verbose         be more verbose (show symbol address, etc)
    -w, --column-widths <width[,width...]>
    -x, --exclude-other   Only display entries with parent-match

Since perf is growing rapidly:

triton:~/tip> L_PREV=0; for v in v2.6.32 v2.6.33 v2.6.34 v2.6.35 v2.6.36 v2.6.37 \
v2.6.38 v2.6.39 v3.0 v3.1 v3.2 v3.3 v3.4 v3.5 v3.6 v3.7 v3.8 v3.9 v3.10 v3.11 \
v3.12 v3.13 v3.14 v3.15 v3.16 v3.17 v3.18 v3.19 v4.0 v4.1 v4.2 master; do printf \ 
"%-10s: " $v; rm -rf tools/perf/; git checkout -f $v >/dev/null 2>&1; L=$(find \
tools/perf/ -type f | xargs cat | wc -l); GROWTH=$((L-L_PREV)); printf \
"%6d  (+%6d) LOC\n" $L $GROWTH; L_PREV=$L; done

v2.6.32   :  27743  (+ 27743) LOC
v2.6.33   :  36676  (+  8933) LOC
v2.6.34   :  41350  (+  4674) LOC
v2.6.35   :  46579  (+  5229) LOC
v2.6.36   :  49377  (+  2798) LOC
v2.6.37   :  51030  (+  1653) LOC
v2.6.38   :  53017  (+  1987) LOC
v2.6.39   :  56141  (+  3124) LOC
v3.0      :  57423  (+  1282) LOC
v3.1      :  58824  (+  1401) LOC
v3.2      :  61403  (+  2579) LOC
v3.3      :  62790  (+  1387) LOC
v3.4      :  66509  (+  3719) LOC
v3.5      :  65967  (+  -542) LOC
v3.6      :  68273  (+  2306) LOC
v3.7      :  74301  (+  6028) LOC
v3.8      :  78281  (+  3980) LOC
v3.9      :  82289  (+  4008) LOC
v3.10     :  84195  (+  1906) LOC
v3.11     :  84536  (+   341) LOC
v3.12     :  90159  (+  5623) LOC
v3.13     :  95169  (+  5010) LOC
v3.14     :  97233  (+  2064) LOC
v3.15     :  98857  (+  1624) LOC
v3.16     : 103760  (+  4903) LOC
v3.17     : 106358  (+  2598) LOC
v3.18     : 109347  (+  2989) LOC
v3.19     : 112867  (+  3520) LOC
v4.0      : 113194  (+   327) LOC
v4.1      : 115992  (+  2798) LOC
v4.2      : 123404  (+  7412) LOC
master    : 143429  (+ 20025) LOC

which is good! ;-) - but managing all the various pieces of documentation will 
become a larger and larger challenge I suspect as time goes on, both for users
and for developers.

Thanks,

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

[toc] | [prev] | [next] | [standalone]

#1254640 — Re: [PATCH 1/3] perf tools: Move callchain help messages to callchain.h

FromArnaldo Carvalho de Melo <acme@kernel.org>
Date2015-10-23 16:30 +0200
SubjectRe: [PATCH 1/3] perf tools: Move callchain help messages to callchain.h
Message-ID<qmLvY-41p-23@gated-at.bofh.it>
In reply to#1254444
Em Fri, Oct 23, 2015 at 11:59:32AM +0200, Ingo Molnar escreveu:
> Btw., another usability detail I noticed yesterday is that when I typed 'perf 
> report -h' I got so much output that I couldn't find the specific option I was 
> looking for, because there's no apparent ordering of the output:
> 
> triton:~/tip> perf report -h 2>&1 | grep -e ' -[a-Z],'
>     -i, --input <file>    input file name
>     -v, --verbose         be more verbose (show symbol address, etc)
>     -D, --dump-raw-trace  dump raw trace in ASCII
SNIP:
>     -C, --cpu <cpu>       list of cpus to profile
>     -I, --show-info       Display extended information about perf.data file
>     -M, --disassembler-style <disassembler style>
>     -b, --branch-stack    use branch records for per branch histogram filling
> 
> Such (alphabetic) ordering would be easier to navigate:
> 
>     -b, --branch-stack    use branch records for per branch histogram filling
>     -c, --comms <comm[,comm...]>
>     -C, --cpu <cpu>       list of cpus to profile

Try the patch below:


From 182ec8d7c105e331884def4f98f56b4f876a4f1d Mon Sep 17 00:00:00 2001
From: Arnaldo Carvalho de Melo <acme@redhat.com>
Date: Fri, 23 Oct 2015 11:23:28 -0300
Subject: [PATCH 1/1] perf tools: Show tool command line options ordered

When asking for a listing of the options, be it using -h or when an
unknown option is passed, order it by one-letter options, then the ones
having just long names.

Suggested-by: Ingo Molnar <mingo@kernel.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Borislav Petkov <bp@suse.de>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Stephane Eranian <eranian@google.com>
Cc: Wang Nan <wangnan0@huawei.com>
Link: http://lkml.kernel.org/n/tip-41qh68t35n4ehrpsuazp1dx8@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
---
 tools/perf/util/parse-options.c | 42 +++++++++++++++++++++++++++++++++++++++++
 1 file changed, 42 insertions(+)

diff --git a/tools/perf/util/parse-options.c b/tools/perf/util/parse-options.c
index 8aa7922397a9..47fb1405df7e 100644
--- a/tools/perf/util/parse-options.c
+++ b/tools/perf/util/parse-options.c
@@ -2,6 +2,7 @@
 #include "parse-options.h"
 #include "cache.h"
 #include "header.h"
+#include <linux/string.h>
 
 #define OPT_SHORT 1
 #define OPT_UNSET 2
@@ -642,9 +643,44 @@ static void print_option_help(const struct option *opts, int full)
 	fprintf(stderr, "%*s%s\n", pad + USAGE_GAP, "", opts->help);
 }
 
+static int option__cmp(const void *va, const void *vb)
+{
+	const struct option *a = va, *b = vb;
+	int sa = tolower(a->short_name), sb = tolower(b->short_name), ret;
+
+	if (sa == 0)
+		sa = 'z' + 1;
+	if (sb == 0)
+		sb = 'z' + 1;
+
+	ret = sa - sb;
+
+	return ret ?: strcmp(a->long_name, b->long_name);
+}
+
+static struct option *options__order(const struct option *opts)
+{
+	int nr_opts = 0;
+	const struct option *o = opts;
+	struct option *ordered;
+
+	for (o = opts; o->type != OPTION_END; o++)
+		++nr_opts;
+
+	ordered = memdup(opts, sizeof(*o) * (nr_opts + 1));
+	if (ordered == NULL)
+		goto out;
+
+	qsort(ordered, nr_opts, sizeof(*o), option__cmp);
+out:
+	return ordered;
+}
+
 int usage_with_options_internal(const char * const *usagestr,
 				const struct option *opts, int full)
 {
+	struct option *ordered;
+
 	if (!usagestr)
 		return PARSE_OPT_HELP;
 
@@ -661,11 +697,17 @@ int usage_with_options_internal(const char * const *usagestr,
 	if (opts->type != OPTION_GROUP)
 		fputc('\n', stderr);
 
+	ordered = options__order(opts);
+	if (ordered)
+		opts = ordered;
+
 	for (  ; opts->type != OPTION_END; opts++)
 		print_option_help(opts, full);
 
 	fputc('\n', stderr);
 
+	free(ordered);
+
 	return PARSE_OPT_HELP;
 }
 
-- 
2.1.0

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

[toc] | [prev] | [next] | [standalone]

#1254707 — Re: [PATCH 1/3] perf tools: Move callchain help messages to callchain.h

FromArnaldo Carvalho de Melo <acme@kernel.org>
Date2015-10-23 18:50 +0200
SubjectRe: [PATCH 1/3] perf tools: Move callchain help messages to callchain.h
Message-ID<qmNHs-762-3@gated-at.bofh.it>
In reply to#1254640
Em Fri, Oct 23, 2015 at 11:27:15AM -0300, Arnaldo Carvalho de Melo escreveu:
> Em Fri, Oct 23, 2015 at 11:59:32AM +0200, Ingo Molnar escreveu:
> > Btw., another usability detail I noticed yesterday is that when I typed 'perf 
> > report -h' I got so much output that I couldn't find the specific option I was 
> > looking for, because there's no apparent ordering of the output:
> > 
> > triton:~/tip> perf report -h 2>&1 | grep -e ' -[a-Z],'
> >     -i, --input <file>    input file name
> >     -v, --verbose         be more verbose (show symbol address, etc)
> >     -D, --dump-raw-trace  dump raw trace in ASCII
> SNIP:
> >     -C, --cpu <cpu>       list of cpus to profile
> >     -I, --show-info       Display extended information about perf.data file
> >     -M, --disassembler-style <disassembler style>
> >     -b, --branch-stack    use branch records for per branch histogram filling
> > 
> > Such (alphabetic) ordering would be easier to navigate:
> > 
> >     -b, --branch-stack    use branch records for per branch histogram filling
> >     -c, --comms <comm[,comm...]>
> >     -C, --cpu <cpu>       list of cpus to profile
> 
> Try the patch below:
> 

I fixed a case where long options could be NULL and implemented this
other patch based on another suggestion you made, which I'll have to fix
to check if opt->long_name is NULL ;-)

From 648b9931e6bccf1d1e1d14474d56b4b687e5fb71 Mon Sep 17 00:00:00 2001
From: Arnaldo Carvalho de Melo <acme@redhat.com>
Date: Fri, 23 Oct 2015 13:27:39 -0300
Subject: [PATCH 1/1] perf tools: Provide help for subset of options

Some tools have a lot of options, so, providing a way to show help just
for some of them may come handy:

  $ perf report -h --tui

   Usage: perf report [<options>]

        --tui             Use the TUI interface

  $ perf report -h --tui --showcpuutilization -b -c

   Usage: perf report [<options>]

    -b, --branch-stack    use branch records for per branch histogram filling
    -c, --comms <comm[,comm...]>
                          only consider symbols in these comms
        --showcpuutilization
                          Show sample percentage for different cpu modes
        --tui             Use the TUI interface

  $

Using it with perf bash completion is also handy, just make sure you
source the needed file:

  $ . ~/git/linux/tools/perf/perf-completion.sh

Then press tab/tab after -- to see a list of options, put them after -h
and only the options chosen will have its help presented:

  $ perf report -h --
  --asm-raw              --demangle-kernel      --group
  --kallsyms             --pretty               --stdio
  --branch-history       --disassembler-style   --gtk
  --max-stack            --showcpuutilization   --symbol-filter
  --branch-stack         --dsos                 --header
  --mem-mode             --show-info            --symbols
  --call-graph           --dump-raw-trace       --header-only
  --modules              --show-nr-samples      --symfs
  --children             --exclude-other        --hide-unresolved
  --objdump              --show-ref-call-graph  --threads
  --column-widths        --fields               --ignore-callees
  --parent               --show-total-period    --tid
  --comms                --field-separator      --input
  --percentage           --socket-filter        --tui
  --cpu                  --force                --inverted
  --percent-limit        --sort                 --verbose
  --demangle             --full-source-path     --itrace
  --pid                  --source               --vmlinux
  $ perf report -h --socket-filter

   Usage: perf report [<options>]

      --socket-filter <n>
                  only show processor socket that match with this filter

Suggested-by: Ingo Molnar <mingo@kernel.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Borislav Petkov <bp@suse.de>
Cc: Brendan Gregg <brendan.d.gregg@gmail.com>
: Chandler Carruth <chandlerc@gmail.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Stephane Eranian <eranian@google.com>
Cc: Wang Nan <wangnan0@huawei.com>
Link: http://lkml.kernel.org/n/tip-83mcdd3wj0379jcgea8w0fxa@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
---
 tools/perf/util/parse-options.c | 42 ++++++++++++++++++++++++++++++++---------
 1 file changed, 33 insertions(+), 9 deletions(-)

diff --git a/tools/perf/util/parse-options.c b/tools/perf/util/parse-options.c
index fb26532d67c3..47ffa90d1ed4 100644
--- a/tools/perf/util/parse-options.c
+++ b/tools/perf/util/parse-options.c
@@ -373,7 +373,8 @@ void parse_options_start(struct parse_opt_ctx_t *ctx,
 }
 
 static int usage_with_options_internal(const char * const *,
-				       const struct option *, int);
+				       const struct option *, int,
+				       struct parse_opt_ctx_t *);
 
 int parse_options_step(struct parse_opt_ctx_t *ctx,
 		       const struct option *options,
@@ -397,8 +398,9 @@ int parse_options_step(struct parse_opt_ctx_t *ctx,
 
 		if (arg[1] != '-') {
 			ctx->opt = ++arg;
-			if (internal_help && *ctx->opt == 'h')
-				return usage_with_options_internal(usagestr, options, 0);
+			if (internal_help && *ctx->opt == 'h') {
+				return usage_with_options_internal(usagestr, options, 0, ctx);
+			}
 			switch (parse_short_opt(ctx, options)) {
 			case -1:
 				return parse_options_usage(usagestr, options, arg, 1);
@@ -413,7 +415,7 @@ int parse_options_step(struct parse_opt_ctx_t *ctx,
 				check_typos(arg, options);
 			while (ctx->opt) {
 				if (internal_help && *ctx->opt == 'h')
-					return usage_with_options_internal(usagestr, options, 0);
+					return usage_with_options_internal(usagestr, options, 0, ctx);
 				arg = ctx->opt;
 				switch (parse_short_opt(ctx, options)) {
 				case -1:
@@ -446,9 +448,9 @@ int parse_options_step(struct parse_opt_ctx_t *ctx,
 
 		arg += 2;
 		if (internal_help && !strcmp(arg, "help-all"))
-			return usage_with_options_internal(usagestr, options, 1);
+			return usage_with_options_internal(usagestr, options, 1, ctx);
 		if (internal_help && !strcmp(arg, "help"))
-			return usage_with_options_internal(usagestr, options, 0);
+			return usage_with_options_internal(usagestr, options, 0, ctx);
 		if (!strcmp(arg, "list-opts"))
 			return PARSE_OPT_LIST_OPTS;
 		if (!strcmp(arg, "list-cmds"))
@@ -682,8 +684,27 @@ out:
 	return ordered;
 }
 
+static bool option__in_argv(const struct option *opt, const struct parse_opt_ctx_t *ctx)
+{
+	int i;
+
+	for (i = 1; i < ctx->argc; ++i) {
+		const char *arg = ctx->argv[i];
+
+		if (arg[0] != '-')
+			continue;
+
+		if (arg[1] == opt->short_name ||
+		    (arg[1] == '-' && strcmp(opt->long_name, arg + 2) == 0))
+			return true;
+	}
+
+	return false;
+}
+
 int usage_with_options_internal(const char * const *usagestr,
-				const struct option *opts, int full)
+				const struct option *opts, int full,
+				struct parse_opt_ctx_t *ctx)
 {
 	struct option *ordered;
 
@@ -707,8 +728,11 @@ int usage_with_options_internal(const char * const *usagestr,
 	if (ordered)
 		opts = ordered;
 
-	for (  ; opts->type != OPTION_END; opts++)
+	for (  ; opts->type != OPTION_END; opts++) {
+		if (ctx && ctx->argc > 1 && !option__in_argv(opts, ctx))
+			continue;
 		print_option_help(opts, full);
+	}
 
 	fputc('\n', stderr);
 
@@ -721,7 +745,7 @@ void usage_with_options(const char * const *usagestr,
 			const struct option *opts)
 {
 	exit_browser(false);
-	usage_with_options_internal(usagestr, opts, 0);
+	usage_with_options_internal(usagestr, opts, 0, NULL);
 	exit(129);
 }
 
-- 
2.1.0

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

[toc] | [prev] | [next] | [standalone]

#1254067 — Re: [PATCH 1/3] perf tools: Move callchain help messages to callchain.h

FromArnaldo Carvalho de Melo <acme@kernel.org>
Date2015-10-22 20:40 +0200
SubjectRe: [PATCH 1/3] perf tools: Move callchain help messages to callchain.h
Message-ID<qmsWm-2kA-1@gated-at.bofh.it>
In reply to#1253502
Em Thu, Oct 22, 2015 at 03:28:48PM +0900, Namhyung Kim escreveu:
> These messages will be used by 'perf top' in the next patch.

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

[toc] | [prev] | [next] | [standalone]

#1254392 — [tip:perf/core] perf tools: Move callchain help messages to callchain.h

Fromtip-bot for Namhyung Kim <tipbot@zytor.com>
Date2015-10-23 10:40 +0200
Subject[tip:perf/core] perf tools: Move callchain help messages to callchain.h
Message-ID<qmG3h-4qs-35@gated-at.bofh.it>
In reply to#1253502
Commit-ID:  21cf62847d29392e51c37460856d3c3c57769c5e
Gitweb:     http://git.kernel.org/tip/21cf62847d29392e51c37460856d3c3c57769c5e
Author:     Namhyung Kim <namhyung@kernel.org>
AuthorDate: Thu, 22 Oct 2015 15:28:48 +0900
Committer:  Arnaldo Carvalho de Melo <acme@redhat.com>
CommitDate: Thu, 22 Oct 2015 15:39:51 -0300

perf tools: Move callchain help messages to callchain.h

These messages will be used by 'perf top' in the next patch.

Signed-off-by: Namhyung Kim <namhyung@kernel.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Borislav Petkov <bp@suse.de>
Cc: Brendan Gregg <brendan.d.gregg@gmail.com>
Cc: Chandler Carruth <chandlerc@gmail.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Stephane Eranian <eranian@google.com>
Cc: Wang Nan <wangnan0@huawei.com>
Link: http://lkml.kernel.org/r/1445495330-25416-1-git-send-email-namhyung@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
---
 tools/perf/builtin-record.c |  8 +-------
 tools/perf/builtin-report.c | 10 +++++++---
 tools/perf/util/callchain.h | 12 ++++++++++++
 3 files changed, 20 insertions(+), 10 deletions(-)

diff --git a/tools/perf/builtin-record.c b/tools/perf/builtin-record.c
index 24ace2f..1a11762 100644
--- a/tools/perf/builtin-record.c
+++ b/tools/perf/builtin-record.c
@@ -1010,13 +1010,7 @@ static struct record record = {
 	},
 };
 
-#define CALLCHAIN_HELP "setup and enables call-graph (stack chain/backtrace) recording: "
-
-#ifdef HAVE_DWARF_UNWIND_SUPPORT
-const char record_callchain_help[] = CALLCHAIN_HELP "fp dwarf lbr";
-#else
-const char record_callchain_help[] = CALLCHAIN_HELP "fp lbr";
-#endif
+const char record_callchain_help[] = CALLCHAIN_RECORD_HELP;
 
 /*
  * XXX Will stay a global variable till we fix builtin-script.c to stop messing
diff --git a/tools/perf/builtin-report.c b/tools/perf/builtin-report.c
index 3b23b25..18a8c52 100644
--- a/tools/perf/builtin-report.c
+++ b/tools/perf/builtin-report.c
@@ -625,6 +625,9 @@ parse_percent_limit(const struct option *opt, const char *str,
 	return 0;
 }
 
+const char report_callchain_help[] = "Display callchains using " CALLCHAIN_REPORT_HELP ". "
+				     "Default: graph,0.5,caller";
+
 int cmd_report(int argc, const char **argv, const char *prefix __maybe_unused)
 {
 	struct perf_session *session;
@@ -699,9 +702,10 @@ int cmd_report(int argc, const char **argv, const char *prefix __maybe_unused)
 		   "regex filter to identify parent, see: '--sort parent'"),
 	OPT_BOOLEAN('x', "exclude-other", &symbol_conf.exclude_other,
 		    "Only display entries with parent-match"),
-	OPT_CALLBACK_DEFAULT('g', "call-graph", &report, "output_type,min_percent[,print_limit],call_order[,branch]",
-		     "Display callchains using output_type (graph, flat, fractal, or none) , min percent threshold, optional print limit, callchain order, key (function or address), add branches. "
-		     "Default: graph,0.5,caller", &report_parse_callchain_opt, callchain_default_opt),
+	OPT_CALLBACK_DEFAULT('g', "call-graph", &report,
+			     "output_type,min_percent[,print_limit],call_order[,branch]",
+			     report_callchain_help, &report_parse_callchain_opt,
+			     callchain_default_opt),
 	OPT_BOOLEAN(0, "children", &symbol_conf.cumulate_callchain,
 		    "Accumulate callchains of children and show total overhead as well"),
 	OPT_INTEGER(0, "max-stack", &report.max_stack,
diff --git a/tools/perf/util/callchain.h b/tools/perf/util/callchain.h
index acee2b3..c9e3a2e 100644
--- a/tools/perf/util/callchain.h
+++ b/tools/perf/util/callchain.h
@@ -7,6 +7,18 @@
 #include "event.h"
 #include "symbol.h"
 
+#define CALLCHAIN_HELP "setup and enables call-graph (stack chain/backtrace) recording: "
+
+#ifdef HAVE_DWARF_UNWIND_SUPPORT
+#define CALLCHAIN_RECORD_HELP  CALLCHAIN_HELP "fp dwarf lbr"
+#else
+#define CALLCHAIN_RECORD_HELP  CALLCHAIN_HELP "fp lbr"
+#endif
+
+#define CALLCHAIN_REPORT_HELP  "output_type (graph, flat, fractal, or none), " \
+	"min percent threshold, optional print limit, callchain order, " \
+	"key (function or address), add branches"
+
 enum perf_call_graph_mode {
 	CALLCHAIN_NONE,
 	CALLCHAIN_FP,
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

[toc] | [prev] | [standalone]

Back to top | Article view | linux.kernel

csiph-web