Messages in this thread | | | Date | Thu, 03 Jan 2008 15:13:07 -0500 | From | David Dillow <> | Subject | Re: [GIT PULL] please pull infiniband.git for-linus |
| |
On Thu, 2008-01-03 at 10:56 -0800, Roland Dreier wrote: > > If we've got time before 2.6.24 final, I'd wait on this a bit. > > ib_srp:srp_remove_work() has them reversed as well, and I'm currently > > tracking down why it oopses when the srp_remove_host() happens before > > the scsi_remove_host(), which is the documented call sequence. > > I think the best thing to do is to merge this (assuming that Linus > gets to it), since it looks quite safe and definitely fixes a crash. > Then if we get to the root cause we can change the order of the calls > if it turns out a different fix is required.
I've made progress on the root cause (posted in another thread), but we need to fix ib_srp.c:srp_remove_work() as well, as I think it will have the same issue. It will only be hit if we cannot reconnect to the target, so it probably doesn't see a lot of use.
I'll send a new patch to you for just the ib_srp.c side. Dave
| |