Maybe it really is finding a parity error? On Thu, Aug 13, 2020 at 4:36 PM Chris Hanson <cmhanson%eschatologist.net@localhost> wrote: > On Aug 13, 2020, at 1:32 PM, Michael Thompson < > michael.99.thompson%gmail.com@localhost> wrote: > > > > > > The MM-6230 manual says "The user must initialize the memory module to > allow the parity generator to store ODD parity otherwise the module will > assert BERR* for any subsequent read cycle." > > > > The first time that NetBSD reads an uninitialized location from this RAM > it will cause a bus error. BetBSD might scan RAM locations to see how much > RAM is available. > > > > You can either write an initialization routine that writes to all memory > locations on the board before NetBSD touches it, or you can move jumper > E1-E2 to E2-E3 to prevent the board from causing a BERR*. > > Indeed, I read this as well, so before installing the board yesterday I > set the jumpers to not generate BERR* on parity error and to auto-recover