perf 工具介绍3

http://blog.chinaunix.net/uid-10540984-id-3854969.html

http://blog.csdn.net/zhangskd/article/details/37902159

[[email protected] ~]# perf record -e cpu-clock ./t1
[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.003 MB perf.data (~137 samples) ]
[[email protected] ~]# perf report
# Samples: 102
#
# Overhead  Command  Shared Object  Symbol
# ........  .......  .............  ......
#
    99.02%       t1  ./t1           [.] longa
     0.98%       t1  [kernel]       [k] do_page_fault

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

usage: perf top [<options>]

    -e, --event <event>   event selector. use ‘perf list‘ to list available events
    -c, --count <n>       event period to sample
    -p, --pid <n>         profile events on existing pid
    -a, --all-cpus        system-wide collection from all CPUs
    -C, --CPU <n>         CPU to profile on
    -k, --vmlinux <file>  vmlinux pathname
    -m, --mmap-pages <n>  number of mmap data pages
    -r, --realtime <n>    collect data with this RT SCHED_FIFO priority
    -d, --delay <n>       number of seconds to delay between refreshes
    -D, --dump-symtab     dump the symbol table used for profiling
    -f, --count-filter <n>
                          only display functions with more events than this
    -g, --group           put the counters into a counter group
    -i, --inherit         child tasks inherit counters
    -s, --sym-annotate <symbol name>
                          symbol to annotate - requires -k option
    -z, --zero            zero history across updates
    -F, --freq <n>        profile at this frequency
    -E, --entries <n>     display this many functions
    -v, --verbose         be more verbose (show counter open errors, etc)