On Wed, 2008-08-27 at 10:14 -0700, Joe Eykholt wrote:
> Rob, this is probably fixed by your upcoming patches.
> 
> The scenario is, fc_rport_reject holds the lport lock while
> calling tt.lport_login, which then calls fc_lport_set_fid,
> which resets the rports.  When it resets the rport to DNS,
> fc_rport() reset grabs the ... wait for it ... lport lock!
> Or tries to.  For a long time.
> 
> If we didn't need to hold the lport lock when calling
> tt.lport_login, then it could be dropped before lport
> calls rport_reset_list.
> 
> Also, if rport didn't need to treat the DNS rport specially,
> we wouldn't run into this problem.
> 
Yeah, I'll try and take care of this with the eventual lport locking
patches. I agree that rport shouldn't consider the DNS rport any
differently than other rports.

_______________________________________________
devel mailing list
[email protected]
http://www.open-fcoe.org/mailman/listinfo/devel

Reply via email to