On Sat, 2009-09-19 at 17:47 +0200, Wolfram Sang wrote: > I can't make it for 2.6.32, but will try to get it queued for 2.6.33. > What about the second FIXME in the driver BTW? It should go away. The timings are correct for slowest type of access (attribute memory if I remember well). At first I thought it would be possible to change the timings dynamically to speedup other memory regions, but a driver may do attribute memory access anytime without warning us, so we have to stick with the slowest value. I will resubmit the patch without the FIXME. -- Maxime
