> -----Original Message----- > From: [email protected] [mailto:[email protected]] > On Behalf Of Nilay Vaish > Sent: Monday, April 11, 2011 2:38 PM > To: M5 Developer List > Subject: Re: [m5-dev] AccessPermission in AbstractEntry
> > > > Could you elaborate on your directory controller question. I suspect > > that you are right and that only one type of directory controller can > > exist in a system, but why is that a problem? > > > > Is it not possible that we have a protocol in which different directory > controllers may behave differently? Ok, I just had a chance to look at the code and I think you a referring to the lack of a directory MACHINETYPE macro in RubySlicc_ComponentMapping.hh. Is that correct? Ideally, there shouldn't be a problem with adding any arbitrarily named controller to Ruby, as long as you incorporate the right component mapping functions into the protocol. However, in practice I suspect it will take some non-trivial amount of modifications to RubySlicc_ComponentMapping.hh. Also you'll need to be careful how Ruby and generate SLICC code uses the auto generated MachineType functions. There may be some tricky issues there. Overall, I can't really provide you a lot of specifics on why the directory MACHINETYPE macro does not exist. There may have been some assumptions behind that that were relevant in GEMS but are no longer valid in gem5. I would grep through the Ruby and generated code for the MachineType functions to fully understand the ramifications. Brad _______________________________________________ m5-dev mailing list [email protected] http://m5sim.org/mailman/listinfo/m5-dev
