Subject: Re: Fault type values for uvm_fault()
To: Ignatios Souvatzis <ignatios@cs.uni-bonn.de>
From: Matthias Drochner <M.Drochner@fz-juelich.de>
List: tech-kern
Date: 03/02/2006 14:13:03
ignatios@cs.uni-bonn.de said:
> What about other architectures? Doesn't any of them pass a valid
> argument?
No, now a 0 is passed everywhere. uvm didn't use this
argument. (And I can't imagine how it could be used,
given that the pmap module is free to have the physical
mapping lag behind the logical.)
As Jason says, it should be removed. I've prepared
a patch which I'll post in a minute.
Unfortunately, we don't save any cycles for now
because uvm_fault() needs one more argument for
the uvm_fault_wire case.
best regards
Matthias