i386/77541: em driver if_oerrors book keeping error

[Craig Leres]

>Number:         77541
>Category:       i386
>Synopsis:       em driver if_oerrors book keeping error
>Confidential:   no
>Severity:       non-critical
>Priority:       low
>Responsible:    freebsd-i386
>State:          open
>Quarter:        
>Keywords:       
>Date-Required:
>Class:          change-request
>Submitter-Id:   current-users
>Arrival-Date:   Tue Feb 15 02:00:41 GMT 2005
>Closed-Date:
>Last-Modified:
>Originator:     Craig Leres
>Release:        FreeBSD 4.10-RELEASE i386
>Organization:
Lawrence Berkeley National Laboratory
>Environment:
>Description:
	It looks like if the chip goes to sleep, em_watchdog() is
	invoked and if_oerrors is incremented. However,
	em_update_stats_counters() overwrites if_oerrors every 2
	seconds with the sum of excessive and late collisions. So
	the output error counter for does not tally watchdog timeouts
	and also can be seen to decrease.

>How-To-Repeat:
>Fix:

	If it's desirable to keep track of watchdog timeouts (I
	believe it is) a simple fix would be to tally watch dog
	timeouts separately and then add them in when calculating
	if_oerrors in em_update_stats_counters().

	An untested, partital suggested fix is appended

==============================================================================

*** if_em.c.virgin	Mon Feb 14 17:49:13 2005
--- if_em.c	Mon Feb 14 17:52:37 2005
***************
*** 780,785 ****
--- 780,786 ----
  	em_init(adapter);
  
  	ifp->if_oerrors++;
+ 	adapter->watchdog_timeout++;
  	return;
  }
  
***************
*** 2897,2903 ****
  	adapter->stats.mpc + adapter->stats.cexterr;
  
  	/* Tx Errors */
! 	ifp->if_oerrors = adapter->stats.ecol + adapter->stats.latecol;
  
  }
  
--- 2898,2905 ----
  	adapter->stats.mpc + adapter->stats.cexterr;
  
  	/* Tx Errors */
! 	ifp->if_oerrors = adapter->watchdog_timeout +
! 	    adapter->stats.ecol + adapter->stats.latecol;
  
  }
  
>Release-Note:
>Audit-Trail:
>Unformatted: