Hmm... Well if you look at io_device.cc getMemoryMode() is used to determine how to do all DMA transactions. So at some level it must be working otherwise we would be hitting asserts in the memory system.
Ali On Mar 1, 2009, at 3:04 AM, Gabe Black wrote: > It was during a page table walk so I'm sure it wasn't an > initialization > ordering issue. I'll look at it again when I revisit FS in the timing > simple CPU. > > Gabe > > Ali Saidi wrote: >> My guess is that you might be checking on it before it's set up. In >> the constructor is probably too early. If you grep for MemoryMode >> you'll see where/how it gets changed. >> >> Ali >> >> On Feb 28, 2009, at 10:49 PM, nathan binkert wrote: >> >> >>> I think that Steve is the only one who both pays attention to the >>> list >>> and might understand what's going on here. >>> >>> Nate >>> >>> On Mon, Feb 23, 2009 at 2:30 AM, Gabe Black <gbl...@eecs.umich.edu> >>> wrote: >>> >>>> I was working on getting the X86_FS working with the timing >>>> simple >>>> CPU as well as it does with atomic, and found that when called in >>>> my >>>> page table walker, sys->getMemoryMode() is very definitely giving >>>> the >>>> wrong answer. I suspect there's something I didn't do quite right >>>> when >>>> setting up the system object, but since part of that is in python I >>>> wasn't sure how it all went together. Why might my system object >>>> think >>>> it's in atomic mode when it's actually in timing? How does it end >>>> up >>>> assuming one mode or the other? >>>> >>>> Gabe >>>> _______________________________________________ >>>> m5-dev mailing list >>>> m5-dev@m5sim.org >>>> http://m5sim.org/mailman/listinfo/m5-dev >>>> >>>> >>>> >>> _______________________________________________ >>> m5-dev mailing list >>> m5-dev@m5sim.org >>> http://m5sim.org/mailman/listinfo/m5-dev >>> >>> >> >> _______________________________________________ >> m5-dev mailing list >> m5-dev@m5sim.org >> http://m5sim.org/mailman/listinfo/m5-dev >> > > _______________________________________________ > m5-dev mailing list > m5-dev@m5sim.org > http://m5sim.org/mailman/listinfo/m5-dev > _______________________________________________ m5-dev mailing list m5-dev@m5sim.org http://m5sim.org/mailman/listinfo/m5-dev
