[PATCH v3] drm/gma500: depend on framebuffer

[Patrik Jakobsson]

On Wed, Apr 20, 2022 at 1:42 AM James Hilliard
<james.hilliard1 at gmail.com> wrote:
>
> On Tue, Apr 12, 2022 at 3:30 AM Patrik Jakobsson
> <patrik.r.jakobsson at gmail.com> wrote:
> >
> > On Tue, Apr 12, 2022 at 3:48 AM James Hilliard
> > <james.hilliard1 at gmail.com> wrote:
> > >
> > > On Mon, Apr 11, 2022 at 3:27 AM Patrik Jakobsson
> > > <patrik.r.jakobsson at gmail.com> wrote:
> > > >
> > > > On Sun, Apr 10, 2022 at 10:05 PM James Hilliard
> > > > <james.hilliard1 at gmail.com> wrote:
> > > > >
> > > > > On Sun, Apr 10, 2022 at 1:52 PM Patrik Jakobsson
> > > > > <patrik.r.jakobsson at gmail.com> wrote:
> > > > > >
> > > > > > On Sun, Apr 10, 2022 at 9:40 PM James Hilliard
> > > > > > <james.hilliard1 at gmail.com> wrote:
> > > > > > >
> > > > > > > On Sun, Apr 10, 2022 at 1:36 PM Patrik Jakobsson
> > > > > > > <patrik.r.jakobsson at gmail.com> wrote:
> > > > > > > >
> > > > > > > > On Sat, Apr 9, 2022 at 6:23 AM James Hilliard <james.hilliard1 at gmail.com> wrote:
> > > > > > > > >
> > > > > > > > > Select the efi framebuffer if efi is enabled.
> > > > > > > > >
> > > > > > > > > This appears to be needed for video output to function correctly.
> > > > > > > > >
> > > > > > > > > Signed-off-by: James Hilliard <james.hilliard1 at gmail.com>
> > > > > > > >
> > > > > > > > Hi James,
> > > > > > > > EFI_FB is its own driver and not needed by gma500 to drive its
> > > > > > > > hardware. What makes you think it's required?
> > > > > > >
> > > > > > > I wasn't getting any HDMI video output without it enabled for some reason,
> > > > > > > I assume it is doing some sort of initialization needed by gma500
> > > > > > > during startup.
> > > > > >
> > > > > > Then it sounds like you might just be using EFI_FB and not gma500. Can
> > > > > > you provide the kernel log with drm.debug=0x1f set on kernel
> > > > > > command-line.
> > > > >
> > > > > Seems efifb loads first and then hands off to gma500
> > > >
> > > > That is how it normally works but efifb shouldn't change the state of
> > > > the currently set mode so shouldn't affect gma500.
> > > > From the logs I can see that you have LVDS (internal panel), HDMI and
> > > > DP (3 displays in total) connected. This sounds wrong. Your version of
> > > > gma500 (Cedarview) doesn't support more than 2 crtcs/pipes. This might
> > > > be a problem.
> > >
> > > Yeah, there's a bug there with the connector status, only DVI-D-1 is actually
> > > connected, I have DP-2 and LVDS-1 turned off in weston.
> >
> > Ok, but are the connectors physically connected to anything?
>
> There is one HDMI cable physically connected to the board
>
> > Regardless of what you do in Weston, the connectors shouldn't be
> > getting modes if they aren't used.
>
> Yeah, it seems there's a bug there, not sure if that's related to the
> output failure when efifb isn't available.
>
> > LVDS might be from VBIOS but I'm
> > not sure where the DP modes come from. It would help if you also
> > provide kernel logs with drm.debug=0x1f when the outputs doesn't work
> > so I have something to compare with.
>
> Log with efifb/fb not enabled in kernel build which causes no
> HDMI output:

After examining the logs I cannot see anything that stands out except
for fbdev emulation not being enabled. Are you sure you just don't
need to compile your kernel with FB support (CONFIG_FB)? If you are
running weston on the fbdev backend then you would obviously not get
any output if FB is not compiled into the kernel.