Message-ID: <CAMeQTsZz1TFUepiH_GufrUQaV4sEYOSs8t+4x89meJ9bt6+tbQ@mail.gmail.com>
Date:   Thu, 21 Apr 2022 21:10:38 +0200
From:   Patrik Jakobsson <patrik.r.jakobsson@...il.com>
To:     James Hilliard <james.hilliard1@...il.com>
Cc:     Thomas Zimmermann <tzimmermann@...e.de>,
        Javier Martinez Canillas <javierm@...hat.com>,
        David Airlie <airlied@...ux.ie>,
        Randy Dunlap <rdunlap@...radead.org>,
        linux-kernel <linux-kernel@...r.kernel.org>,
        dri-devel <dri-devel@...ts.freedesktop.org>
Subject: Re: [PATCH v3] drm/gma500: depend on framebuffer

On Thu, Apr 21, 2022 at 7:48 PM James Hilliard
<james.hilliard1@...il.com> wrote:
>
> On Thu, Apr 21, 2022 at 8:20 AM Thomas Zimmermann <tzimmermann@...e.de> wrote:
> >
> > Hi
> >
> > Am 21.04.22 um 14:54 schrieb Patrik Jakobsson:
> > > On Thu, Apr 21, 2022 at 2:47 PM Javier Martinez Canillas
> > > <javierm@...hat.com> wrote:
> > >>
> > >> Hello Patrik,
> > >>
> > >> On 4/21/22 14:39, Patrik Jakobsson wrote:
> > >>> On Thu, Apr 21, 2022 at 1:49 PM Thomas Zimmermann <tzimmermann@...e.de> wrote:
> > >>>>
> > >>>> Hi
> > >>>>
> > >>>> Am 09.04.22 um 06:23 schrieb James Hilliard:
> > >>>>> 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@...il.com>
> > >>>>> ---
> > >>>>> Changes v2 -> v3:
> > >>>>>     - select EFI_FB instead of depending on it
> > >>>>> Changes v1 -> v2:
> > >>>>>     - use depends instead of select
> > >>>>> ---
> > >>>>>    drivers/gpu/drm/gma500/Kconfig | 2 ++
> > >>>>>    1 file changed, 2 insertions(+)
> > >>>>>
> > >>>>> diff --git a/drivers/gpu/drm/gma500/Kconfig b/drivers/gpu/drm/gma500/Kconfig
> > >>>>> index 0cff20265f97..a422fa84d53b 100644
> > >>>>> --- a/drivers/gpu/drm/gma500/Kconfig
> > >>>>> +++ b/drivers/gpu/drm/gma500/Kconfig
> > >>>>> @@ -2,11 +2,13 @@
> > >>>>>    config DRM_GMA500
> > >>>>>        tristate "Intel GMA500/600/3600/3650 KMS Framebuffer"
> > >>>>>        depends on DRM && PCI && X86 && MMU
> > >>>>> +     depends on FB
> > >>>>
> > >>>> Why do we need FB here? Framebuffer support should be hidden by DRM's
> > >>>> fbdev helpers.
> > >>>
> > >>> It is not needed but gives him video output since it enables the drm
> > >>> fbdev emulation.
> > >>>
> > >>
> > >> I'm not sure to understand this. Shouldn't depend on DRM_FBDEV_EMULATION then?
> >
> > It has to be selected separately in .config.
> >
> > James, make sure you have DRM_FBDEV_EMULATION enabled in your kernel's
> > .config file if you want a console. however, Weston should work without
> > it.  How do you start weston without a console?
>
> Well it seems to be needed for weston to work over HDMI, I'm running
> weston as a systemd service.
>
> >
> > >
> > > No, it shouldn't depend on any FBDEV stuff since it's not actually
> > > required. It just happens to help in this case since weston + fbdev
> > > backend works but not weston with drm backend (or whatever config
> > > James have set).
> > >
> > >>
> > >>> I looked some more at the logs and it seems weston doesn't work on his
> > >>> system without the fbdev backend. So the question is why weston isn't
> > >>> working without fbdev? Perhaps this is just a Weston configuration
> > >>> issue?
> > >>>
> > >>
> > >> But is weston using the fbdev emulated by DRM or the one registered by
> > >> efifb? I thought that the latter from what was mentioned in this thread.
> > >
> > > It's using drm fbdev emulation with gma500 so EFIFB has nothing to do
> > > with this. I believe it was just simply incorrectly reported. If I'm
> > > correct then "depends on FB" is what makes video output work for
> > > James.
> >
> > There's an fbdev backend for weston, but the provided logs don't look as
> > if it's using this.  Maybe the connector's modes are never probed
> > correctly without fbdev.
> >
> > James, could you provide a logfile for weston? (i.e., 'weston
> > --log=weston.txt').
>
> Here's the log with output working over HDMI(DVI-D-1):
> Date: 2022-04-21 UTC
> [17:43:11.726] weston 10.0.0
>                https://wayland.freedesktop.org
>                Bug reports to:
> https://gitlab.freedesktop.org/wayland/weston/issues/
>                Build: 10.0.0
> [17:43:11.727] Command line: /bin/weston
> --config=/etc/xdg/weston/weston.ini --log=/srv/weston.txt -Swayland-1
> [17:43:11.727] OS: Linux, 5.17.3, #5 SMP PREEMPT Thu Apr 21 11:20:47
> MDT 2022, x86_64
> [17:43:11.727] Flight recorder: enabled
> [17:43:11.727] Using config file '/etc/xdg/weston/weston.ini'
> [17:43:11.727] Output repaint window is 7 ms maximum.
> [17:43:11.727] Loading module '/usr/lib/libweston-10/drm-backend.so'
> [17:43:11.743] initializing drm backend
> [17:43:11.743] Trying logind launcher...
> [17:43:11.743] logind: cannot find systemd session for uid: 0 -61
> [17:43:11.743] logind: cannot setup systemd-logind helper error: (No
> data available), using legacy fallback
> [17:43:11.744] Trying weston_launch launcher...
> [17:43:11.744] could not get launcher fd from env
> [17:43:11.744] Trying direct launcher...
> [17:43:11.748] using /dev/dri/card0
> [17:43:11.748] DRM: does not support atomic modesetting
> [17:43:11.748] DRM: does not support GBM modifiers
> [17:43:11.748] DRM: supports picture aspect ratio
> [17:43:11.765] event2  - Power Button: is tagged by udev as: Keyboard
> [17:43:11.765] event2  - Power Button: device is a keyboard
> [17:43:11.771] event3  - Video Bus: is tagged by udev as: Keyboard
> [17:43:11.772] event3  - Video Bus: device is a keyboard
> [17:43:11.777] event0  - Power Button: is tagged by udev as: Keyboard
> [17:43:11.778] event0  - Power Button: device is a keyboard
> [17:43:11.785] event1  - Sleep Button: is tagged by udev as: Keyboard
> [17:43:11.787] event1  - Sleep Button: device is a keyboard
> [17:43:11.799] event4  - PiKVM Composite KVM Device: is tagged by udev
> as: Keyboard
> [17:43:11.799] event4  - PiKVM Composite KVM Device: device is a keyboard
> [17:43:11.859] event5  - PiKVM Composite KVM Device: is tagged by udev as: Mouse
> [17:43:11.861] event5  - PiKVM Composite KVM Device: device is a pointer
> [17:43:11.891] libinput: configuring device "Power Button".
> [17:43:11.891] libinput: configuring device "Video Bus".
> [17:43:11.891] libinput: configuring device "Power Button".
> [17:43:11.892] libinput: configuring device "Sleep Button".
> [17:43:11.892] libinput: configuring device "PiKVM Composite KVM Device".
> [17:43:11.892] libinput: configuring device "PiKVM Composite KVM Device".
> [17:43:11.892] input device event5 has no enabled output associated
> (none named), skipping calibration for now.
> [17:43:11.907] DRM: head 'VGA-1' updated, connector 37 is disconnected.
> [17:43:11.907] DRM: head 'VGA-1' found, connector 37 is disconnected.
> [17:43:11.912] DRM: head 'LVDS-1' updated, connector 39 is connected,
> EDID make 'unknown', model 'unknown', serial 'unknown'
> [17:43:11.912] DRM: head 'LVDS-1' found, connector 39 is connected,
> EDID make 'unknown', model 'unknown', serial 'unknown'
> [17:43:12.336] DRM: head 'DVI-D-1' updated, connector 41 is connected,
> EDID make 'TSB', model 'Toshiba-H2C', serial '2290649088'
> [17:43:12.336] DRM: head 'DVI-D-1' found, connector 41 is connected,
> EDID make 'TSB', model 'Toshiba-H2C', serial '2290649088'
> [17:43:12.337] DRM: head 'DP-1' updated, connector 43 is disconnected.
> [17:43:12.337] DRM: head 'DP-1' found, connector 43 is disconnected.
> [17:43:12.369] DRM: head 'DVI-D-2' updated, connector 47 is disconnected.
> [17:43:12.369] DRM: head 'DVI-D-2' found, connector 47 is disconnected.
> [17:43:12.425] DRM: head 'DP-2' updated, connector 49 is connected,
> EDID make 'CHR', model '1024x768', serial '880'
> [17:43:12.425] DRM: head 'DP-2' found, connector 49 is connected, EDID
> make 'CHR', model '1024x768', serial '880'
> [17:43:12.426] Registered plugin API 'weston_drm_output_api_v1' of size 24
> [17:43:12.426] Registered plugin API
> 'weston_drm_virtual_output_api_v1' of size 48
> [17:43:12.426] Color manager: no-op
> [17:43:12.426] Output 'DVI-D-1' using color profile: built-in default
> sRGB SDR profile
> [17:43:12.426] DRM: output DVI-D-1 uses shadow framebuffer.
> [17:43:12.426] Output DVI-D-1 (crtc 32) video modes:
>                1280x720@...0, preferred, current, 74.2 MHz
>                1920x1080@...0 16:9, 148.5 MHz
>                1920x1080@...0 16:9, 74.2 MHz
>                1920x1080@...0 16:9, 74.2 MHz
>                1920x1080@...0 16:9, 74.2 MHz
>                1280x1024@...0, 135.0 MHz
>                1440x900@...9, 106.5 MHz
>                1280x720@...0 16:9, 74.2 MHz
>                1280x720@...9 16:9, 74.2 MHz
>                1280x720@...0 16:9, 74.2 MHz
>                1280x720@...0 16:9, 74.2 MHz
>                1280x720@...0 16:9, 74.2 MHz
>                1280x720@...0 16:9, 59.4 MHz
>                1024x768@...1, 75.0 MHz
>                1024x768@...0, 65.0 MHz
>                800x600@...0, 49.5 MHz
>                800x600@...2, 50.0 MHz
>                800x600@...3, 40.0 MHz
>                800x600@...2, 36.0 MHz
>                720x576@...0 16:9, 27.0 MHz
>                720x576@...0 4:3, 27.0 MHz
>                720x480@...0 4:3, 27.0 MHz
>                720x480@...0 16:9, 27.0 MHz
>                720x480@...9, 27.0 MHz
>                720x480@...9, 27.0 MHz
>                720x480@...9 16:9, 27.0 MHz
>                720x480@...9 4:3, 27.0 MHz
>                640x480@...0, 31.5 MHz
>                640x480@...8, 31.5 MHz
>                640x480@...7, 30.2 MHz
>                640x480@...0 4:3, 25.2 MHz
>                640x480@...9, 25.2 MHz
>                640x480@...9 4:3, 25.2 MHz
> [17:43:12.427] associating input device event2 with output DVI-D-1
> (none by udev)
> [17:43:12.427] associating input device event3 with output DVI-D-1
> (none by udev)
> [17:43:12.427] associating input device event0 with output DVI-D-1
> (none by udev)
> [17:43:12.427] associating input device event1 with output DVI-D-1
> (none by udev)
> [17:43:12.427] associating input device event4 with output DVI-D-1
> (none by udev)
> [17:43:12.427] associating input device event5 with output DVI-D-1
> (none by udev)
> [17:43:12.428] Output 'DVI-D-1' enabled with head(s) DVI-D-1
> [17:43:12.428] Compositor capabilities:
>                arbitrary surface rotation: yes
>                screen capture uses y-flip: no
>                cursor planes: yes
>                arbitrary resolutions: no
>                view mask clipping: yes
>                explicit sync: no
>                color operations: no
>                presentation clock: CLOCK_MONOTONIC, id 1
>                presentation clock resolution: 0.000000001 s
> [17:43:12.428] Loading module '/usr/lib/weston/kiosk-shell.so'
> [17:43:12.429] Note: support for the deprecated wl_shell interface is
> disabled. If a legacy client still needs it, it can be re-enabled by
> passing -Ddeprecated-wl-shell=true to Meson when building Weston.
> [17:43:12.429] Loading module '/usr/lib/weston/systemd-notify.so'

Ok, so my assumption that the working config was using westons fbdev
backend was wrong.

Looking further at the kernel logs I can see that in the non-working
case the HDMI connector gets crtc-1 instead of crtc-0. The pipe
selection for the port seems correct. Not sure what is going on.

Powered by blists - more mailing lists