On 09/14/2010 12:00 PM, Michael Tokarev wrote:
# strace -p 23020 -p 23023 -c Process 23020 attached - interrupt to quit Process 23023 attached - interrupt to quit ^CProcess 23020 detached Process 23023 detached % time seconds usecs/call calls errors syscall ------ ----------- ----------- --------- --------- ---------------- 98.98 23.462158 15569 1507 ioctl 0.64 0.151725 37 4063 5 futex 0.11 0.025051 0 86932 select 0.09 0.021188 0 408836 clock_gettime 0.07 0.016543 0 271933 136484 read 0.04 0.008958 0 181118 gettimeofday 0.02 0.005756 0 101190 write 0.02 0.005506 0 89180 timer_settime 0.02 0.004754 0 96271 timer_gettime 0.01 0.002682 0 50595 rt_sigaction 0.00 0.000000 0 1 rt_sigpending 0.00 0.000000 0 1 1 rt_sigtimedwait ------ ----------- ----------- --------- --------- ---------------- 100.00 23.704321 1291627 136490 total
Well, well. Translating to rates, we have 63.6 ioctls/sec but 17k clock_gettime, 11k read() and 4k writes per second. The device model has gone haywire, somehow without any trace in perf top.
Let's see what those reads and writes are about. Please strace a short segment again, without -c, and use lsof to see which fds are referenced in the read and write calls.
-- error compiling committee.c: too many arguments to function -- To unsubscribe from this list: send the line "unsubscribe kvm" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
