Geoff Wing wrote:

>On Thursday 2006-06-08 11:09 +1000, Frank Kardel output:
>:Please test NetBSD-current so we can smoke out bugs early.
>
>One more tiny problem in FAST_IPSEC build.
>time_second is defined here in <netipsec/ipsec_osdep.h> with NetBSD
>and not __HAVE_TIMECOUNTERS
>
>Regards,
>Geoff
>
>--- sys/netipsec/ipsec.c.org	2006-04-12 17:37:59.000000000 +1000
>+++ sys/netipsec/ipsec.c	2006-06-10 19:16:29.000000000 +1000
>@@ -311,7 +311,7 @@
> 		 */
> 	}
> 
>-	pcbsp->sp_cache[dir].cachesp->lastused = mono_time.tv_sec;
>+	pcbsp->sp_cache[dir].cachesp->lastused = time_second;
> 	pcbsp->sp_cache[dir].cachesp->refcnt++;
> 	KEYDEBUG(KEYDEBUG_IPSEC_STAMP,
> 		printf("DP ipsec_checkpcbcache cause refcnt++:%d SP:%p\n",
>
>  
>
Thanks for detecting that. Generally that code should be checked whether
UTC time is the correct time scale to measure life times and timeouts.

(fixed in 1.23)

Frank