Re: [RFC][PATCH] nbd driver for 2.5+: fix locking issues with ioctl UI

[Lou Langholtz <ldl@aros.net>]

Lou Langholtz wrote:

> . . . On the other hand I've been thinking that I might be able to 
> take advantage of the irq locked condition imposed by the 
> q->queue_lock and just use nbd_lock to replace q->queue_lock then. Al 
> and Andrew seem to have a much deeper understanding though for 
> spinlocking though so I'll defer to there comments on this idea (of 
> replacing lo->queue_lock by use of nbd_lock). This has the added 
> attraction of already having nbd_lock locked when in do_nbd_request.. . .

Typo! Above should have read "just use nbd_lock to replace 
lo->queue_lock" (another spinlock_t per nbd_device). Anyways... would 
using the one nbd_lock to also protect the lo->queue_list work better 
than using the queue_lock per nbd_device I'm wondering. According to the 
prior discusions about spinlocks this should be better. I don't have a 
picture right now of wether that even works or not. Gotta run though, 
thanks!