David Nusinow, le Fri 15 Jul 2005 15:04:05 -0400, a écrit : > To be honest, I'd rather not mess with this file unless we have to. The > simple fix in Redhat's bugzilla looks a lot less invasive, and it seems to > fix all the reported problems so far, so I'd rather go with that unless > upstream says otherwise. I'm wondering: why isn't the mmio base _itself_ qualified by volatile ? It definitely _is_ intrinsicly volatile. Regards, Samuel