fntimes.stp - Show Functions Taking Longer Than Usual
The fntimes.stp script monitors the execution time history of a given
function family (assumed non-recursive). Each time (beyond a warmup
interval) is then compared to the historical maximum. If it exceeds
a certain threshold (250%), a message is printed.
$ stap fntimes.stp 'kernel.function("sys_*")' -c "sleep 7"
function SyS_ioctl well over maximum time (32 vs 12)
function SyS_open well over maximum time (2832 vs 188)
function SyS_newfstat well over maximum time (5 vs 2)
function SyS_mmap_pgoff well over maximum time (282 vs 12)
function SyS_mmap well over maximum time (286 vs 15)
function SyS_access well over maximum time (35 vs 11)
function SyS_access well over maximum time (92 vs 35)
function SyS_mprotect well over maximum time (44 vs 13)
function SyS_futex well over maximum time (9753 vs 3260)
function SyS_newstat well over maximum time (4882 vs 133)
function SyS_recvmsg well over maximum time (607 vs 44)
function SyS_futex well over maximum time (55896 vs 9753)
function SyS_futex well over maximum time (500089 vs 55927)
function SyS_poll well over maximum time (832629 vs 53308)
Getting no output is not a bad thing; it means that means that functions
did not exceed the threshold.
This output was gathered by running ls repeatedly in another terminal
a few seconds after the script started.
y~or5J ={Eeu磝Qk ᯘG{?+]ן?wM3X^歌>{7پK>on\jy R g/=fOroNVv~Y+ NGuÝHWyw[eQʨSb> >}Gmx[o[<{Ϯ_qF vM IENDB`
