amd64/92412: rcp.rstatd reports bogus packets/per/second info

[Sean McNeil]

On Jan 30, 2006, at 11:02 AM, hotlips Internet admin wrote:

> Thus saith Sean McNeil:
> | On Sun, 2006-01-29 at 07:30 +0000, hotlips Internet admin wrote:
> | > The following reply was made to PR amd64/92412; it has been  
> noted by GNATS.
> | >
> | > From: hotlips Internet admin <hostmaster at GTS.NET>
> | > To: kris at obsecurity.org (Kris Kennaway)
> | > Cc: bug-followup at FreeBSD.org
> | > Subject: Re: amd64/92412: rcp.rstatd reports bogus packets/per/ 
> second info
> | > Date: Sun, 29 Jan 2006 02:25:33 -0500 (EST)
> | >
> | >  Thus saith Kris Kennaway:
> | >  | On Fri, Jan 27, 2006 at 11:59:15AM -0500, hotlips Internet  
> admin wrote:
> | >  |
> | >  | > | > >Description:
> | >  | > | > packets/second value reported by rpc.rstatd to remote  
> monitor hovers around
> | >  | > | > 8000 or so with odd downward spikes approx every 90  
> seconds (it's not at
> | >  | > | all related to actual interface traffic)
> | >  | > | >
> | >  | > | > >How-To-Repeat:
> | >  | > | > keep displaying rpc.rstatd data from 6.0 system
> | >  | > |
> | >  | > | How are you using rpc.rstatd and rup?  I don't see a way  
> to make rup
> | >  | > | display "packets/second", it only gives uptime and load  
> average:
> | >  | > |
> | >  | > | # rup
> | >  | > | fbsd-amd64.isc.  10:02am  up   4 days,  14:00,  load  
> average: 2.00 2.00 2.00
> | >  | >
> | >  | > 	Solaris perfmeter, actually.
> | >  |
> | >  | Do you know how I can query this on FreeBSD?
> | >
> | >
> | >  	oop, wrong (older) diff - use this one below...
> |
> | The diff would be more readable if you only include useful  
> changes and
> | eliminate difference caused solely from white space.  For  
> instance, with
> | diff -ub.
>
>
> 	rup is also broken, sigh - patches below to account for
> 	8-btye timevals
>
>
> -- 
> Bruce Becker			+1 416 410 0879
> GTS Network Administration	Toronto, Ont.
> Email:	hostmaster at gts.net
>
>  --------- 8< --------- 8< --------- 8< --------- 8< --------- 8<  
> ---------
>
> --- rup.c.orig	Sat May 21 05:55:07 2005
> +++ rup.c	Mon Jan 30 13:55:34 2006
> @@ -93,6 +93,7 @@
>  static bool_t
>  rstat_reply(caddr_t replyp, struct sockaddr_in *raddrp)
>  {
> +	long longtime;
>  	struct tm *tmp_time;
>  	struct tm host_time;
>  	struct tm host_uptime;
> @@ -118,12 +119,14 @@
>
>  	printf("%-*s\t", HOST_WIDTH, host);
>
> -	tmp_time = localtime((time_t *)&host_stat->curtime.tv_sec);
> +	longtime = host_stat->curtime.tv_sec;
> +	tmp_time = localtime((time_t *)&longtime);
>  	host_time = *tmp_time;
>


Just curious, but why not declare a variable as time_t instead of  
long?  Seems like that would be more correct.


>  	host_stat->curtime.tv_sec -= host_stat->boottime.tv_sec;
>
> -	tmp_time = gmtime((time_t *)&host_stat->curtime.tv_sec);
> +	longtime = host_stat->curtime.tv_sec;
> +	tmp_time = gmtime((time_t *)&longtime);
>  	host_uptime = *tmp_time;
>
>  	#define updays (host_stat->curtime.tv_sec  / 86400)
>
>