Em Thu, Jul 09, 2020 at 12:45:48AM +0800, Changbin Du escreveu: > Here is a glance of all ftrace functions with this serias: > > $ sudo perf ftrace -h > Usage: perf ftrace [<options>] [<command>] > or: perf ftrace [<options>] -- <command> [<options>] > > -a, --all-cpus system-wide collection from all CPUs > -C, --cpu <cpu> list of cpus to monitor > -d, --delay <n> ms to wait before starting tracing after program > start > -D, --graph-depth <n> > Max depth for function graph tracer (This option is > deprecated) > -F, --funcs Show available functions to filter > -G, --graph-funcs <func> > trace given functions using function_graph tracer > -g, --nograph-funcs <func> > Set nograph filter on given functions > -m, --buffer-size <size> > size of per cpu buffer > -N, --notrace-funcs <func> > do not trace given functions > -p, --pid <pid> trace on existing process id > -t, --tid <tid> trace on existing thread id (exclusive to --pid) > -T, --trace-funcs <func> > trace given functions using function tracer > -t, --tracer <tracer> > tracer to use: function or function_graph (This > option is deprecated) > -v, --verbose be more verbose > --func-opts <options> > function tracer options, available options: > call-graph,irq-info > --graph-opts <options> > graph tracer options, available options: > nosleep-time,noirqs,verbose,thresh=<n>,depth=<n> > --inherit trace children processes > > v3: > o add --func-opts and --graph-opts to set tracer specific options. > o support units as a suffix for option '-m/--buffer-size'.
Thanks for working on this! -d/--delay: [acme@quaco perf]$ perf record -h delay Usage: perf record [<options>] [<command>] or: perf record [<options>] -- <command> [<options>] -D, --delay <n> ms to wait before starting measurement after program start [acme@quaco perf]$ [acme@quaco perf]$ perf stat -h --delay Usage: perf stat [<options>] [<command>] -D, --delay <n> ms to wait before starting measurement after program start [acme@quaco perf]$ [acme@quaco perf]$ perf trace -h -D Usage: perf trace [<options>] [<command>] or: perf trace [<options>] -- <command> [<options>] or: perf trace record [<options>] [<command>] or: perf trace record [<options>] -- <command> [<options>] -D, --delay <n> ms to wait before starting measurement after program start Perhaps we can move --graph-depth to --graph-opts and make -D/--delay be just like with 'perf record', 'perf stat' and 'perf trace'? Will be super nice if one can go from: perf stat --delay 100 workload to see the default set of counters to: perf ftrace --delay 100 workload to get the ftrace output for that. Ditto seeing the syscalls strace-like with: perf trace --delay 100 workload Regards, - Arnaldo