Re: LOR: tcpinp and tcp

Hi,

Gavin Atkinson wrote on Wed, May 12, 2004 at 08:59:14PM +0100:
[..]
> It looks like this LOR has been reported once before, in early January. It
> seems to still be an issue. This is with top-of-tree CURRENT. It's not a
> known lock according to the LOR page.
> 
> lock order reversal
>  1st 0xc2d7a2ac inp (tcpinp) _at_ /usr/src/sys/netinet/tcp_input.c:717
>  2nd 0xc08b230c tcp (tcp) _at_ /usr/src/sys/netinet/tcp_usrreq.c:616
> Stack backtrace:
[..]

I just came across this one, as well. It is now mentioned on
the Zabbadoz LOR-page, but still as "unknown".

Here is my stack:
[..]
> lock order reversal
>  1st 0xc6840630 inp (tcpinp) _at_ /usr/src/sys/netinet/tcp_input.c:717
>  2nd 0xc076d98c tcp (tcp) _at_ /usr/src/sys/netinet/tcp_usrreq.c:616
> Stack backtrace:
> backtrace(0,ffffffff,c0747858,c0747bc8,c0712e1c) at backtrace+0x12
> witness_checkorder(c076d98c,9,c06e30f1,268) at witness_checkorder+0x593
> _mtx_lock_flags(c076d98c,0,c06e30f1,268) at _mtx_lock_flags+0x67
> tcp_usr_rcvd(c687e780,80) at tcp_usr_rcvd+0x1b
> soreceive(c687e780,d7e61b2c,d7e61b38,d7e61b30,0) at soreceive+0x865
> nfsrv_rcv(c687e780,c6241b80,4) at nfsrv_rcv+0x83
> sowakeup(c687e780,c687e7cc) at sowakeup+0x71
> tcp_input(c199ec00,14,0,14,812a9f83) at tcp_input+0x12ba
> ip_input(c199ec00) at ip_input+0x83a
> netisr_processqueue(c076c858,c4392580,c198ae00,d7e61d1c,c05355a8) at netisr_processqueue+0x6e
> swi_net(0) at swi_net+0x85
> ithread_loop(c198ae00,d7e61d48,c198ae00,c0535404,0) at ithread_loop+0x1a4
> fork_exit(c0535404,c198ae00,d7e61d48) at fork_exit+0xa8
> fork_trampoline() at fork_trampoline+0x8
> --- trap 0x1, eip = 0, esp = 0xd7e61d7c, ebp = 0 ---
[..]

It did not cause a panic and appeared to do no harm. Maybe this
turns out a false positive as well. Can anybody confirm this?

Best regards,
 Daniel
-- 
IRCnet: Mr-Spock                                    - Eddie would go! -
 Daniel Lang * dl_at_leo.org * +49 89 289 18532 * http://www.leo.org/~dl/