On Thu, 7 Feb 2019 16:41:58 +0000 Chris Wilson <chris@xxxxxxxxxxxxxxxxxx> wrote: > Quoting Bob Paauwe (2019-02-07 16:29:53) > > With the address range being specified for each platform, we can use > > that instead of the .ppgtt enum to handle the differences between > > 3 level and 4 level PPGTT. In most cases, we really only care if the > > platform supports PPGTT or not. Because of this, we can now remove > > the HAS_FULL_PPGTT macro and the device info ppgtt field. > > > > Aliasing PPGTT used by GEN 6 is a bit of an exception. For those cases, > > it makes just as much sense to check if we're running on GEN 6 as it > > does to check a device info flag. > > > > v2: Reword the commit message to make it correct wrt aliasing ppgtt (Chris) > > v3: Rebase on current drm-tip > > The point of adding the type into the device_info was that it was > included in the error state so that I didn't have to remember which w/a > applied to which gen. gen6 has full-ppgtt, we can't enabled it as no one > has solved how to make it work. > -Chris OK, I had assumed that the point was to control the code flow so I'm not sure how best to proceed. This patch doesn't really change any program logic so we can drop it and keep both the ppgtt type and size in device info. Alternately, I could add the ppgtt size to error state output. But I'm not sure that really provides what you're looking for wrt gen6. If you have a better idea or something specific you'd like to see me do with this, I could use some direction. Thanks, Bob -- -- Bob Paauwe Bob.J.Paauwe@xxxxxxxxx IOTG / PED Software Organization Intel Corp. Folsom, CA (916) 356-6193 _______________________________________________ Intel-gfx mailing list Intel-gfx@xxxxxxxxxxxxxxxxxxxxx https://lists.freedesktop.org/mailman/listinfo/intel-gfx
