Experimental Physics and
Industrial Control System

Ralph Lange <[email protected]> · Sat, 12 May 2007 00:04:34 +0200

Hi,

I think I found something that smells like a bug in the vxWorks "local"

clock (or in libCom/osi/os/vxWorks/iocClock).

When the TIMEZONE environment variable gets set through

configure/CONFIG_SITE_ENV in base, everything works as expected.

When I set the TIMEZONE variable to the exact same value from the

startup script (at the very beginning), the iocClock (i.e. the "local"

IOC time when calling "date" in the VxWorks shell) always returns

  2038/01/19 04:14:08.nnnn

where only the split second is changing, as if the "seconds since the

epoch" are constant and only the nanoseconds count up. The same applies

to the "local" time generated by a stringin record with "Soft Timestamp"

support: Correct string if TIMEZONE originates from CONFIG_SITE_ENV,

January 2038 if set in the startup script.

The timestamps that CA delivers cross network are coherent and correct.

only local conversion seems to be affected.

The behaviour is exactly the same when using apsEvent 1.1 (the 3.14

version of the TS... module) in soft slave mode instead of the iocClock

from libCom.

I don't see anything obviously stupid in libCom/osi/os/vxWorks/iocClock.c.

System: VxWorks 5.4.2, EPICS 3.14.8.2, MV2100

Is someone able to test and/or confirm this behaviour?

Cheers,
Ralph

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System