[Intel-gfx] [PATCH v2 0/7] drm: drm_fb_helper cleanup.

Daniel Vetter daniel at ffwll.ch
Mon Mar 2 21:45:08 UTC 2020


On Mon, Mar 02, 2020 at 10:43:19PM +0100, Daniel Vetter wrote:
> On Mon, Mar 02, 2020 at 06:21:23PM +0000, Emil Velikov wrote:
> > Hi Pankaj,
> > 
> > On Mon, 2 Mar 2020 at 16:33, Pankaj Bharadiya
> > <pankaj.laxminarayan.bharadiya at intel.com> wrote:
> > >
> > > This series addresses below drm_fb_helper tasks from
> > > Documentation/gpu/todo.rst.
> > >
> > > - The max connector argument for drm_fb_helper_init() isn't used
> > >   anymore and can be removed.
> > >
> > > - The helper doesn't keep an array of connectors anymore so these can
> > >   be removed: drm_fb_helper_single_add_all_connectors(),
> > >   drm_fb_helper_add_one_connector() and
> > >   drm_fb_helper_remove_one_connector().
> > >
> > > Changes since v1:
> > >    - Squashed warning fixes into the patch that introduced the
> > >      warnings (into 5/7) (Laurent)
> > >    - Fixed reflow in in 9/9 (Laurent)
> > >
> > For the future, include the changelog in the respective patches. This
> > makes it easier for reviewers...
> > Plus you're already changing the commit - might as well mention what/why :-)
> > 
> > Also do include the R-B, Acked-by, other tags accumulated up-to that
> > point, when sending new revision.
> 
> Yup, can you pls resend that entire pile with all the ack/review tags from
> the earlier versions included? If you don't do that you waste reviewers
> time. Another one is that resend right away also kinda wastes peoples
> time, because there's a much bigger chance that someone will review the
> old version, instead of your new one. Better wait of at least 1-2 days or
> so.

Hm just noticed that people are still reviewing v1. I'd say lets forget
about this v2 here, wait 1-2 days, and then resend with everything
combined. Hopefully not too many will re-review v2 here and waste time on
stuff that's reviewed already. Resending within hours is really not good
on mailing lists (with merge requests or whatever it doesn't matter,
because everyone always looks at the latest version).
-Daniel

> 
> > That said, if you're interested in further cleaning this up, one can
> > cleanup the drm_dp_mst_topology_cbs hooks.
> > In particular ::register_connector is identical across the board -
> > create a helper function using it directly in core, killing the hook.
> > 
> > While for ::destroy_connector - there's some amdgpu specific code in
> > there... which I'm not sure if it should stay or not.
> > To be on the save side - create a helper which will be called for
> > drivers where the hook is !=NULL (aka everyone but amdgpu).
> 
> Yeah that stuff looks fishy. Smells a bit like old mst code developed
> before Lyude fixed the refcounting for real, it seems to manually shut
> down stuff that should be cleaned up with normal code paths (modeset
> and/or final kref_put on the connector).
> -Daniel
> -- 
> Daniel Vetter
> Software Engineer, Intel Corporation
> http://blog.ffwll.ch

-- 
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch


More information about the dri-devel mailing list