[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: ssnwait routine

This archive contains answers to questions sent to Unidata support through mid-2025. Note that the archive is no longer being updated. We provide the archive for reference; many of the answers presented here remain technically correct, even if somewhat outdated. For the most up-to-date information on the use of NSF Unidata software and data services, please consult the Software Documentation first.

  • Subject: Re: ssnwait routine
  • Date: Fri, 16 Dec 2005 14:53:13 -0500
Hi guys - I've managed to reproduce Christian's error w/ a minimalist C
program ... but it *only* appears on x86_64 systems -- identical code
runs w/o warnings w/ FC4 on i386.  I'll stick in the mod that you
suggest and take if for another spin on the 64bit machine ... will keep
you posted :)
ron

On Fri, 2005-12-16 at 12:47 -0700, Steve Chiswell wrote:
> Ron and Kevin,
> 
> I'll ammend my previous support reply about the ssnwait.f routine which
> calls:
> CALL CSSLEEP  ( nanosec, ier )
> 
> Here, the nanosec value is defined in GEMPRM.PRM
> PARAMETER       ( NANOSEC = 59999999 )
> 
> By passing this parameter value to cssleep as "long *nanosec",
> the gcc compiler is treating the quantity as a modifiable quantity
> which seems to confuse the optimizer in certain cases.
> 
> Even though the cssleep routine does not modify the value, it could-
> which is a runtime error seen in several Garp routines recently,
> so I've modified the ssnwait.f routine as follows:
> 
>         nsecval = nanosec
>         CALL CSSLEEP  ( nsecval, ier )
> 
> I haven't seen Christian's error here, but he may want to try
> that possibility to see if that is related or not.
> 
> Steve Chiswell
> Unidata User Support
>