Re: [SSI-devel] An amusing bug. But what on earth should it be doing.

[John H. <john@Calva.COM>]

John Hughes wrote:
> Investigating a kernel BUG report, unlocking a lock that was not locked,
> bug #1944781 I find the following code in cluster/ssi/ipc/ipcshm_svr.c:
>   
[...]
> So if id is > 0 it's going to call spin_unlock.  But it didn't call
> spin_lock, so *BUG*.
>
> Why should (despite it's name) ripc_drop_locks call ipc_drop_locks?
>   
Well, it should, but it only wants ipc_drop_locks to do the "up ()"
operation, not the "ipc_unlock()" so it should call ipc_drop_locks with
id "0".