A pair of LORs

[Antoine Brodin]

"Bjoern A. Zeeb" <bzeeb-lists at lists.zabbadoz.net> wrote:
> On Mon, 14 Mar 2005, Antoine Brodin wrote:
> 
> > Derek Tattersall <dlt at mebtel.net> wrote:
> > > This happened on a 6.0 current system supped and built this
> > > morning. I didn't see this recorded on Herr Zeeb's page, so
> > > perhaps somebody has some clues to offer?
> > [snip]
> > > lock order reversal
> > > 1st 0xc26f1300 so_rcv (so_rcv) @
> > > /usr/src/sys/kern/uipc_socket.c:1256 2nd 0xc23a96c4 skc0 (network
> > > driver) @ /usr/src/sys/pci/if_sk.c:1112
> >
> > For the second LOR, you can try the patch available at:
> > http://sources.zabbadoz.net/freebsd/patchset/if_sk.c-HEAD-20050303-01.diff
> 
> if you are going to try that please use the [1-4]of4 patches from
> same directory and apply in ascending order.
> 
> But I doubt it will help with this LOR (but testing would be good
> anyway;)
> 
> An interesting thing to know would be if you have sk0 and sk1 (dual
> port) ?

I think it can definitively help: with the patch, sk_jfree uses a leaf
mutex and no longer uses the network driver mutex.

Cheers,

Antoine