kern/78957: time counter per process stops (syscall: getrusage)
Jakub Kruszona-Zawadzki
acid at gemius.pl
Thu Mar 17 12:50:03 PST 2005
>Number: 78957
>Category: kern
>Synopsis: time counter per process stops (syscall: getrusage)
>Confidential: no
>Severity: non-critical
>Priority: medium
>Responsible: freebsd-bugs
>State: open
>Quarter:
>Keywords:
>Date-Required:
>Class: sw-bug
>Submitter-Id: current-users
>Arrival-Date: Thu Mar 17 20:50:02 GMT 2005
>Closed-Date:
>Last-Modified:
>Originator: Jakub Kruszona-Zawadzki
>Release: FreeBSD 4.x - 5.x
>Organization:
Gemius SA
>Environment:
FreeBSD adnet.hit.gemius.pl 4.10-RELEASE FreeBSD 4.10-RELEASE #0: Thu Oct 7 13:48:42 CEST 2004 hitcol at adocean.hit.gemius.pl:/usr/src/sys/compile/ADNET i386
(but it occurs on every FreeBSD 4.x and 5.x)
>Description:
When a process is running for a long time (several days) time counter per process stops on value:
ru_utime.tv_sec:305221
ru_utime.tv_usec:322735
>How-To-Repeat:
write something like this:
void main() {
struct rusage rus;
while (1) {
.. do something that uses the processor for a few minutes - loop with some calculations ...
getrusage(RUSAGE_SELF,&rus);
printf("ru_utime.tv_sec:%ld | ru_utime.tv_usec:%ld",rus.ru_utime.tv_sec,rus.ru_utime.tv_usec);
}
}
compile and run then wait about 305000s (about 3.5 days)
>Fix:
No idea - may be something is wrongly calculated in kernel ???
>Release-Note:
>Audit-Trail:
>Unformatted:
More information about the freebsd-bugs
mailing list