modesetting and fbdev

modesetting and fbdev

[Jorge Luis Zapata Muga]

Hi all, I've seen some patches and discussions about modesetting into
the kernel and fb devices, what is the relation or plans with that
beside the pros and cons? keep the fb api? some kind of merge between
both?
Regards.

-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2008.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/

Re: modesetting and fbdev

[Jesse Barnes]

On Wednesday 30 January 2008 11:55:19 am Jorge Luis Zapata Muga wrote:
> Hi all, I've seen some patches and discussions about modesetting into
> the kernel and fb devices, what is the relation or plans with that
> beside the pros and cons? keep the fb api? some kind of merge between
> both?

The DRM modesetting code preserves compatibility with the fb interfaces, so if 
you write a DRM based modesetting driver, fb applications can run on top 
unmodified.  Also, the existing fb stuff won't go away (though some 
maintainers may choose to replace their fb drivers with DRM modesetting based 
ones).  Which type of driver to write just depends on what your hw is capable 
of and whether you feel the slightly higher complexity of a DRM based driver 
is justified (it probably will be if you ever plan on adding 3d support to 
your stack or if you want to exploit the fancy output control that the DRM 
based API provides).

Hope that helps.

Jesse

-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2008.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/

Re: modesetting and fbdev

[Jorge Luis Zapata Muga]

On Jan 31, 2008 5:55 AM, Jesse Barnes <jbarnes@virtuousgeek.org> wrote:
>
> On Wednesday 30 January 2008 11:55:19 am Jorge Luis Zapata Muga wrote:
> > Hi all, I've seen some patches and discussions about modesetting into
> > the kernel and fb devices, what is the relation or plans with that
> > beside the pros and cons? keep the fb api? some kind of merge between
> > both?
>
> The DRM modesetting code preserves compatibility with the fb interfaces, so if
> you write a DRM based modesetting driver, fb applications can run on top
> unmodified.  Also, the existing fb stuff won't go away (though some
> maintainers may choose to replace their fb drivers with DRM modesetting based
> ones).  Which type of driver to write just depends on what your hw is capable
> of and whether you feel the slightly higher complexity of a DRM based driver
> is justified (it probably will be if you ever plan on adding 3d support to
> your stack or if you want to exploit the fancy output control that the DRM
> based API provides).
>
> Hope that helps.
>

Yes it does, thanks for explaining ... my concerns are about multiple
framebuffers on one chip, from what i see you reference a framebuffer
== output (correct me if im wrong), and i thought that maybe this
upper controller (drm) could take the control of hw with multiple
graphics layers i.e one framebuffer for every gfx layer, like
controlling the compositing of them, beside the memory controller/3d
thing. Is this in consideration?

Where is the main development taken? on the modesetting-101 branch?

Thanks!

> Jesse
>

-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2008.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/

Re: modesetting and fbdev

[Jesse Barnes]

On Thursday, January 31, 2008 3:58 pm Jorge Luis Zapata Muga wrote:
> On Jan 31, 2008 5:55 AM, Jesse Barnes <jbarnes@virtuousgeek.org> 
wrote:
> > On Wednesday 30 January 2008 11:55:19 am Jorge Luis Zapata Muga 
wrote:
> > > Hi all, I've seen some patches and discussions about modesetting
> > > into the kernel and fb devices, what is the relation or plans
> > > with that beside the pros and cons? keep the fb api? some kind of
> > > merge between both?
> >
> > The DRM modesetting code preserves compatibility with the fb
> > interfaces, so if you write a DRM based modesetting driver, fb
> > applications can run on top unmodified.  Also, the existing fb
> > stuff won't go away (though some maintainers may choose to replace
> > their fb drivers with DRM modesetting based ones).  Which type of
> > driver to write just depends on what your hw is capable of and
> > whether you feel the slightly higher complexity of a DRM based
> > driver is justified (it probably will be if you ever plan on adding
> > 3d support to your stack or if you want to exploit the fancy output
> > control that the DRM based API provides).
> >
> > Hope that helps.
>
> Yes it does, thanks for explaining ... my concerns are about multiple
> framebuffers on one chip, from what i see you reference a framebuffer
> == output (correct me if im wrong),

Yeah, that's wrong. :)  It supports multiple memory objects, which you 
can bind to CRTCs, which connect to outputs, arbitrarily.

> and i thought that maybe this 
> upper controller (drm) could take the control of hw with multiple
> graphics layers i.e one framebuffer for every gfx layer, like
> controlling the compositing of them, beside the memory controller/3d
> thing. Is this in consideration?

Yes, definitely.  One of the design goals for these new parts of the DRM 
layer was to fully support output reconfiguration and routing, so the 
system should be as flexible as the underlying hardware allows.

> Where is the main development taken? on the modesetting-101 branch?

Yep, that's the main tree right now.

Jesse

-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2008.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/