yamt@mwd.biglobe.ne.jp writes:

>>Number:         30154
>>Category:       kern
>>Synopsis:       tcp_close locking botch
>>Confidential:   no
>>Severity:       serious
>>Priority:       medium
>>Responsible:    kern-bug-people
>>State:          open
>>Class:          sw-bug
>>Submitter-Id:   net
>>Arrival-Date:   Fri May 06 14:15:00 +0000 2005
>>Originator:     YAMAMOTO Takashi <yamt@mwd.biglobe.ne.jp>
>>Release:        NetBSD 3.99.3
>>Organization:
>
>>Environment:
> 	
> 	
> System: NetBSD kaeru 3.99.3 NetBSD 3.99.3 (build.kaeru.xen) #78: Wed May 4 23:21:47 JST 2005 takashi@kaeru:/home/takashi/work/kernel/build.kaeru.xen i386

My system is GENERIC.MP, CVSed May 3 or 4.

> Architecture: i386
> Machine: i386
>>Description:
> 	tcp_close can be called from interrupt context.
> 	however, it isn't safe.
>
> 	eg. tcp_input -> tcp_close -> in_pcbdetach
> 	    -> sofree -> chgsbsize -> uid_find
>>How-To-Repeat:
> 	code inspection.

Router, ftp, web, nat server.

Just wait from 30 min till 2 days uptime. No panic, just hangs.
It needs hard reset.

Excerpt from ddb:
simple_lock: locking against myself
lock:0xc07ff628 currently at kern_resource.c:907 on CPU 1
last locked: kern_resource.c:904
last unlocked: kern_resource.c:907

and backtrace is the same as mentioned above.

>>Fix:
> 	
>
>>Unformatted:
>  	
>  	
>
>

-- 
= Alexey G. Khramkov (agkhram) @ Infotech <agkhram{at}mercdev{dot}com> =
= GPG fingerprint : 944D 0C8B 343B 6C8D 50A1  061A E2DA 3E11 7765 6B47 =