On Mon, Nov 3, 2008 at 3:15 PM, Wolfram Sang <[EMAIL PROTECTED]> wrote: > On Mon, Nov 03, 2008 at 03:57:09PM -0600, Matt Sealey wrote: > >> > Optionally you can further >>> reduce impact by checking if CONFIG_PPC_MPC5200_BUGFIX is defined. >> >> I would much prefer this. > > I submitted a patch to enable pipelining on a MPC5200B recently. It was > disabled because of a bug in the MPC5200. The first version of this > patch used MPC5200_BUGFIX and it was mentioned, that some people might > want to run the same kernel on both kind of processors. So, the patch > that went mainline checks for the PVR. Maybe we should stick to this > here, too?
Of the two solutions: 1. Run-time selection must be done. 2. Compile-time removal of the bug fix path can also be done to lessen runtime impact for kernels never run on older chips My view is that #1 is non-negotiable. #2 is a nice to have, but if it doesn't incur any cost when disabled at runtime then I don't care. g. -- Grant Likely, B.Sc., P.Eng. Secret Lab Technologies Ltd. _______________________________________________ Linuxppc-dev mailing list Linuxppc-dev@ozlabs.org https://ozlabs.org/mailman/listinfo/linuxppc-dev
