Re: [PATCH] drm/i915/bdw: remove preliminary_hw_support flag from BDW

[Daniel Vetter <daniel@ffwll.ch>]

On Tue, Jan 14, 2014 at 02:05:45PM -0800, Ben Widawsky wrote:
> On Tue, Jan 14, 2014 at 11:04:16AM -0800, Jesse Barnes wrote:
> > It ought to work ok in 3.14.  We have some fun stuff coming after that,
> > but all the basics are in place now and seem relatively stable.
> > 
> > Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
> > ---
> >  drivers/gpu/drm/i915/i915_drv.c | 2 --
> >  1 file changed, 2 deletions(-)
> > 
> > diff --git a/drivers/gpu/drm/i915/i915_drv.c b/drivers/gpu/drm/i915/i915_drv.c
> > index 61fb9fc..df4d350 100644
> > --- a/drivers/gpu/drm/i915/i915_drv.c
> > +++ b/drivers/gpu/drm/i915/i915_drv.c
> > @@ -342,7 +342,6 @@ static const struct intel_device_info intel_haswell_m_info = {
> >  };
> >  
> >  static const struct intel_device_info intel_broadwell_d_info = {
> > -	.is_preliminary = 1,
> >  	.gen = 8, .num_pipes = 3,
> >  	.need_gfx_hws = 1, .has_hotplug = 1,
> >  	.ring_mask = RENDER_RING | BSD_RING | BLT_RING | VEBOX_RING,
> > @@ -351,7 +350,6 @@ static const struct intel_device_info intel_broadwell_d_info = {
> >  };
> >  
> >  static const struct intel_device_info intel_broadwell_m_info = {
> > -	.is_preliminary = 1,
> >  	.gen = 8, .is_mobile = 1, .num_pipes = 3,
> >  	.need_gfx_hws = 1, .has_hotplug = 1,
> >  	.ring_mask = RENDER_RING | BSD_RING | BLT_RING | VEBOX_RING,
> 
> I had been hoping to at least get the following in or checked out before
> marking as not preliminary:
> 
> semaphores - patch series has one regression, but I would have liked it
> merged and turned off
> 
> RC6 well tested - I am not sure what the state of our HW is in this
> regard. I was told the HW we have doesn't support it, but I haven't
> tried.
> 
> RPS - see RC6
> 
> 4G GGTT - I guess this is currently blocked on PPGTT, which is
> unfortunate. I can live without this for non-preilm.
> 
> 
> I've been trying my best to keep the bug count low, and it has been
> pretty good thanks to you and Paulo - so I think it should be safe
> enough to call the HW supported, even with my gripes above. So while I
> prefer to get the stuff above in/tested, if you want to move ahead with
> this, it's
> 
> Acked by: Ben Widawsky <ben@bwidawsk.net>

The target for dropping the preliminary hw support tag is "works at least
as good as vesa". That means
- modesetting works with the boot config (broken hotplug in corner cases
  is meh, but imo ok)
- suspend/resume doesn't blow up
- userspace isn't in such a bad shape to take down the entire machine

All the things you're listing are goodies on top (of course we want them
for the shipping driver, but that's not the criterion here at all). We
have the preliminary hw support thing since at least on hsw and on byt
we've had entire clases of outputs that didn't work at all (DP on hsw and
mipi on byt).

The goal here is that users can always at least install a distro (and then
upgrade to a decent kernel) with whatever gets burned into the images,
without the need to resort to stuff like adding nomodeset.

Patch merged for 3.14, thanks.
-Daniel
-- 
Daniel Vetter
Software Engineer, Intel Corporation
+41 (0) 79 365 57 48 - http://blog.ffwll.ch