> Date: Fri, 26 Oct 2018 09:31:56 +1100 > From: Jonathan Gray <j...@jsg.id.au> > > On Thu, Oct 25, 2018 at 08:12:13PM +0200, Mark Kettenis wrote: > > Found the issue. The kernel advertises support for the I915_MMAP_WC > > option, but we don't actually implement that functonality correctly > > :(. As a result we create a cachable mapping instead of a > > write-combining (uncached) one. On hardware generations that aren't > > cache coherent this means that the hardware may see stale memory > > contents. So the GPU starts executing some random garbage which is > > very likely to result in a hang. > > > > The diff below stops advertising I915_MMAP_WC support. This is a > > workaround. I'm looking into implementing I915_MMAP_WC properly, but > > that may require a bit of uvm surgery. > > Thanks for tracking this down, ok jsg@ if you also make > i915_gem_mmap_ioctl() error on args->flags & I915_MMAP_WC.
I committed the diff as-is last night since Theo wanted to get a usuable snap out because of the Xorg thing. My plan is to have a proper I915_MMAP_WC implementation ready soon anyway.