Re: [PATCH 1/1] perf:Adding --list-opts to usage string

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

Em Wed, Oct 14, 2015 at 10:29:05AM +0800, Yunlong Song escreveu:
> On 2015/10/13 23:24, Ramkumar Ramachandra wrote:
> > Arnaldo Carvalho de Melo wrote:
> >> Em Thu, Oct 08, 2015 at 10:59:50AM +0530, Sriram Raghunathan escreveu:
> >>> Minor change, adding --list-opts to usage string. So that it is
> >>> visible to the user on running perf --help. or just perf
> >>> from command line.
> >>
> >> Ramkumar, Yunlong, are you ok with this?

> > Not sure I understand the motivation, but I suppose it can't hurt to
> > show this detail?

> Agree with Ramkumar, --list-opts is redundant due to the existing [OPTIONS] in
> the perf_usage_string[].

I see, thinking about it now, it seems that this is because 'perf -h'
behaves differently from other tools, i.e.:

$ perf -h

 usage: perf [--version] [--help] [OPTIONS] COMMAND [ARGS]

 The most commonly used perf commands are:
   annotate        Read perf.data (created by perf record) and display annotated code
   archive         Create archive with object files with build-ids found in perf.data file
   bench           General framework for benchmark suites
   buildid-cache   Manage build-id cache.
   buildid-list    List the buildids in a perf.data file
<SNIP>
   test            Runs sanity tests.
   timechart       Tool to visualize total system behavior during a workload
   top             System profiling tool.
   trace           strace inspired tool
   probe           Define new dynamic tracepoints

 See 'perf help COMMAND' for more information on a specific command.

--------------------------

While:

$ perf stat -h

 usage: perf stat [<options>] [<command>]

    -T, --transaction     hardware transaction statistics
    -e, --event <event>   event selector. use 'perf list' to list available events
        --filter <filter>
                          event filter
    -i, --no-inherit      child tasks do not inherit counters
    -p, --pid <pid>       stat events on existing process id
    -t, --tid <tid>       stat events on existing thread id
    -a, --all-cpus        system-wide collection from all CPUs
<SNIP>
    -I, --interval-print <n>
                          print counts at regular interval in ms (>= 10)
        --per-socket      aggregate counts per processor socket
        --per-core        aggregate counts per physical processor core
        --per-thread      aggregate counts per thread
    -D, --delay <n>       ms to wait before starting measurement after program start

--------------------------

One doesn't show what options can be used, the other does, so there is
an inconsistency, this and the fact that 'perf -h' outputs to stdout,
'perf stat -h' and the other builtins output to stderr. I think all
should output to stdout, just like 'ls --help', what do you think?

- Arnaldo
 
> >>>
> >>>  const char perf_usage_string[] =
> >>> -     "perf [--version] [--help] [OPTIONS] COMMAND [ARGS]";
> >>> +     "perf [--version] [--help] [--list-opts] [OPTIONS] COMMAND [ARGS]";
> 
> -- 
> Thanks,
> Yunlong Song