Hannes Reinecke wrote:
> Why do we have to unload the modules here?
> Of course it's nice and proper in theory, but you get
> a nice crash when running multipath over iscsi.
> As iscsi is stopped prior to multipathing it's okay
> to shutdown the daemon (as the connection will still be
> functioning).
> But as there's still a multipath map on these devices
> (ie bd_claim is set) the sysfs / driver core objects
> won't be removed but having invalid function pointers
> instead.

I do not think this will happen in current iscsi modules will it? If the 
multipath device has the iscsi device in the map, then doesn't it still 
have a reference to the scsi_device? If it does then the iscsi_tcp 
module will have a refcount and the rmmod will fail. If the iscsi_tcp 
still has a refcount greater than zero then so does libiscsi and so will 
scsi_transport_iscsi, so those modules will not be removed.

In past modules we had that bug where we did not set the module 
reference in the scsi_host_template, so you (or some code like 
multipath) could be using a device then you could log a session off or 
rmmod iscsi_tcp and then you hit the problem you describe. You need this 
patch:

http://git.kernel.org/?p=linux/kernel/git/jejb/scsi-misc-2.6.git;a=commitdiff;h=7974392c0b0d4e7a2a17ca3597d51a29b9841aa5

--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups 
"open-iscsi" group.
To post to this group, send email to open-iscsi@googlegroups.com
To unsubscribe from this group, send email to [EMAIL PROTECTED]
For more options, visit this group at http://groups.google.com/group/open-iscsi
-~----------~----~----~----~------~----~------~--~---

Reply via email to