There is parameter defer_registration that is used in the BaseCPU class.
It is set to true in case the cpu will not come in to action immediately.
We can pass this on to the interrupt object and in case it is true, not do
anything with the pio port.
--
Nilay
On Wed, 22 Feb 2012, Ali Saidi wrote:
What probably needs to happen to fix this is the creation of the
params in BaseCPU and the instatiation of objects like the interrupt
controllers need to be done is seperate steps. The latter needs to only
be called on CPUs that are going to be executing code, not on every CPU.
Ali
On 22.02.2012 14:58, Ali Saidi wrote:
Hmm.. That makes some
sense. The issue is that the interrupt object
that the timing cpu is
instantiating is never used. When the switch
happens the timing cpu
gets attached to the one that the atomic cpu was
using. Something that
would probably work is assigning the timing cpu's
interrupt object to
None.
Ali
On 22.02.2012 14:31, Nilay Vaish
wrote:
Hi
As you might have noticed there have been several
emails on the
users
mailing list about the interrupts object not
being
connected to anything.
Here is the situation --
suppose one
wants to switch from Atomic CPU to
Interrupts objects
are
created for the two CPUs. Since this object is a
Links:
------
[1]
mailto:[email protected]
[2] http://m5sim.org/mailman/listinfo/gem5-dev
_______________________________________________
gem5-dev mailing list
[email protected]
http://m5sim.org/mailman/listinfo/gem5-dev
_______________________________________________
gem5-dev mailing list
[email protected]
http://m5sim.org/mailman/listinfo/gem5-dev
