On Thu, 2009-02-26 at 11:13 -0500, John W. Linville wrote:
> > Now I just have to get around to fixing it, it's proving to be harder
> > than I thought, and I don't have much time right now.
>
> I suppose I could just drop that one. IIRC there was another patch
> from you that depended upon that one as well?
Ok, so I looked into this, and it turns out that it's harder to fix than
I thought. Fixing the genl code to call the pre_dumpit at the right
place is hard and requires netlink (not genl) changes. I'm not in a
position to do such changes right now, so I would suggest to revert it,
and for now take the easy way of restructuring nl80211 to take, in each
operation, the rtnl before any other locks instead of trying to do this
generically.
Luis, would you be up to that? Basically go from
mutex_lock(&cfg80211_mtx)
...
rtnl_lock()
...
rtnl_unlock()
...
mutex_unlock(&cfg80211_mtx)
to
rtnl_lock()
...
mutex_lock(&cfg80211_mtx)
...
mutex_unlock(&cfg80211_mtx)
...
rtnl_unlock()
everywhere.
johannes
Attachment:
signature.asc
Description: This is a digitally signed message part
