On Fri, Sep 08, 2006 at 10:23:36PM +0200, Francois Romieu wrote: > > I suspect it's a chip bug. I rechecked with I/O space, and that > > works okay, so this artifact (bug) only manifests itself when you > > do the upper write in MMIO space. > > > > Are there any plans to switch r8169 to the iomap API? Would you > > take a patch if I'd write one? > > Given the current state of the r8169 driver, I do not see a lot of > benefit from the iomap() API in itself. > > It could make the switch to I/O read/write easier for strange bugs > like your but I have an epidermic defiance against I/O ops (much too > synchronizing for me: people forget that MMIO will post). I may > change my mind if bugs start poping up like mushrooms but we are > hopefully not there yet. > > An ordered write with a big sign in front of it to comment the issue > is good enough for me. > > Don't hesitate to protest if you think that I need a clue.
What you say makes sense -- in my case it would have been useful to have a knob to switch the driver to use I/O ops (since that is what the vendor driver uses, and the vendor driver works), but bugs like these are generally rare anyway. and so the added benefit isn't too big. OK. cheers, Lennert - To unsubscribe from this list: send the line "unsubscribe netdev" in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html