Subject: RE: sleeping on a userland address
To: None <pjh@speakeasy.net>
From: Bill Studenmund <wrstuden@netbsd.org>
List: tech-kern
Date: 05/22/2002 14:16:51
On Wed, 22 May 2002, Peter J. Hunter wrote:
> The sleep code won't care as it just treats the ident as a key to a hash
> table, casting it as a u_long. However, you'll lose the unique
> qualities of the key by using user space addresses (which can be shared
> between apps). Therefore, you run the risk of being errantly woken up
> and errantly waking up others.
Also, some ports use a seperate kernel address space, so you might in
principle end up waking up other waiters waiting on a kernel address. :-)
Take care,
Bill