Re: [PATCH v5] drm/radeon: fix internal display on iMac11, 1 (RV770/DCE3.1)

On Thu, Aug 20, 2026 at 2:59 PM Gilles Risch <[email protected]> wrote:
>
> Good afternoon,
>
> no problem, please find attached my complete dmesg output after a fresh
> reboot. I don't see a bit difference, except that my card is connected
> to a different PCI bus.
>
> I don’t pass any kernel parameters via GRUB and I don’t use a xorg.conf
> configuration file either. Have you connected an additional, external
> display?
> Kind regards,
> Gilles
>
> On 20.08.26 15:24, Alex Deucher wrote:
> > On Thu, Aug 20, 2026 at 8:50 AM Erik Gregg <[email protected]> wrote:
> >> On Mon, 1 Jun 2026 23:19:32 +0200, Gilles Risch <[email protected]> wrote:
> >>> The Apple iMac11,1 (27-inch, Late 2009) uses a Mobility Radeon HD 4850
> >>> (RV770/DCE3.1) with a 2560x1440 internal panel on an internal
> >>> DisplayPort path. Without this fix the display stays dark under KMS.
> >>>
> >>> This machine suffers from the same issue as iMac10,1 and iMac11,2:
> >>> Apple routes the internal display through Link B of the DIG encoder
> >>> instead of Link A. Add iMac11,1 to the existing DMI quirk and move
> >>> the Apple-specific encoder assignment into its own block, independent
> >>> of the DCE version check.
> >>>
> >>> Additionally, the 2560x1440 panel requires RADEON_PLL_USE_FRAC_FB_DIV
> >>> and ATOM_ENCODER_CMD_DP_VIDEO_ON, limited to iMac11,1 via dmi_match()
> >>> to avoid affecting other boards.
> >>>
> >>> Signed-off-by: Gilles Risch <[email protected]>
> >> I tested this patch on top of 7.1.7-arch1 on a Late 2009 27" iMac11,1
> >> with the Mobility Radeon HD 4850. With nomodeset the machine boots with
> >> a working (efifb) console. On vanilla 7.1.7 without nomodeset the panel
> >> stays completely black (backlight off) and no radeon init is visible in
> >> dmesg over SSH.
> >>
> >> With the patch and without nomodeset, radeon initializes normally per
> >> dmesg, but at the fbcon modeset (~4s) the panel goes blank and then
> >> shows flashing horizontal lines that start at the top and sweep down at
> >> regular intervals. It never recovers. The logs below are from that case,
> >> gathered by logging in blind on the console and running commands blind.
> >>
> >> I tried radeon.dpm=0, radeon.audio=0, video=eDP-1:2560x1440@60 and
> >> radeon.modeset=1, with no change.
> > I wonder if different variants of the board are wired up differently.
> > @Gilles Risch, can you attach your dmesg output as well?  Maybe there
> > are different panels on different variants?

@Erik Gregg Can you narrow down which hunk(s) if this patch are
problematic?  If I had to guess, I'd say it's this hunk:

+       if (dmi_match(DMI_PRODUCT_NAME, "iMac10,1") ||
+           dmi_match(DMI_PRODUCT_NAME, "iMac11,1") ||
+           dmi_match(DMI_PRODUCT_NAME, "iMac11,2")) {
+               enc_idx = (dig->linkb) ? 1 : 0;
+               goto assigned;
+       }

Alex

> >
> > Alex
> >
> >> Renderer: AMD RV770 (DRM 2.51.0 / 7.1.7-arch1-1-imac), Mesa 26.1.6
> >>
> >> [    3.028962] radeon 0000:02:00.0: fence driver on ring 3 uses gpu addr 0x0000000020000c0c
> >> [    3.029207] radeon 0000:02:00.0: fence driver on ring 5 uses gpu addr 0x0000000000056230
> >> [    3.030250] radeon 0000:02:00.0: radeon: using MSI.
> >> [    3.030275] [drm] radeon: irq initialized.
> >> [    3.076796] [drm] ring test on 0 succeeded in 1 usecs
> >> [    3.076803] [drm] ring test on 3 succeeded in 2 usecs
> >> [    3.252855] radeon 0000:02:00.0: [drm] ring test on 5 succeeded in 1 usecs
> >> [    3.252861] radeon 0000:02:00.0: [drm] UVD initialized successfully.
> >> [    3.253264] [drm] ib test on ring 0 succeeded in 0 usecs
> >> [    3.253283] [drm] ib test on ring 3 succeeded in 0 usecs
> >> [    3.906199] radeon 0000:02:00.0: [drm] ib test on ring 5 succeeded
> >> [    3.906669] [drm] radeon atom DIG backlight initialized
> >> [    3.906672] radeon 0000:02:00.0: [drm] Radeon Display Connectors
> >> [    3.906674] radeon 0000:02:00.0: [drm] Connector 0:
> >> [    3.906676] radeon 0000:02:00.0: [drm]   DP-1
> >> [    3.906677] radeon 0000:02:00.0: [drm]   HPD2
> >> [    3.906679] radeon 0000:02:00.0: [drm]   DDC: 0x7e60 0x7e60 0x7e64 0x7e64 0x7e68 0x7e68 0x7e6c 0x7e6c
> >> [    3.906682] radeon 0000:02:00.0: [drm]   Encoders:
> >> [    3.906683] radeon 0000:02:00.0: [drm]     DFP1: INTERNAL_UNIPHY
> >> [    3.906685] radeon 0000:02:00.0: [drm] Connector 1:
> >> [    3.906687] radeon 0000:02:00.0: [drm]   eDP-1
> >> [    3.906688] radeon 0000:02:00.0: [drm]   HPD3
> >> [    3.906689] radeon 0000:02:00.0: [drm]   DDC: 0x7e20 0x7e20 0x7e24 0x7e24 0x7e28 0x7e28 0x7e2c 0x7e2c
> >> [    3.906692] radeon 0000:02:00.0: [drm]   Encoders:
> >> [    3.906693] radeon 0000:02:00.0: [drm]     LCD1: INTERNAL_UNIPHY
> >> [    3.906695] radeon 0000:02:00.0: [drm] Connector 2:
> >> [    3.906697] radeon 0000:02:00.0: [drm]   VGA-1
> >> [    3.906698] radeon 0000:02:00.0: [drm]   DDC: 0x7e40 0x7e40 0x7e44 0x7e44 0x7e48 0x7e48 0x7e4c 0x7e4c
> >> [    3.906701] radeon 0000:02:00.0: [drm]   Encoders:
> >> [    3.906702] radeon 0000:02:00.0: [drm]     CRT1: INTERNAL_KLDSCP_DAC1
> >> [    3.915213] [drm] Initialized radeon 2.51.0 for 0000:02:00.0 on minor 0
> >> [    3.988184] [drm] fb mappable at 0xC0347000
> >> [    3.988187] [drm] vram apper at 0xC0000000
> >> [    3.988188] [drm] size 14745600
> >> [    3.988189] [drm] fb depth is 24
> >> [    3.988190] [drm]    pitch is 10240
> >> [    3.988263] fbcon: radeondrmfb (fb0) is primary device
> >> [    4.026022] radeon 0000:02:00.0: [drm] fb0: radeondrmfb frame buffer device
> >> [    5.366673] systemd[1]: Load Kernel Module drm skipped, unmet condition check ConditionKernelModuleLoaded=!drm
> >> [    8.464118] snd_hda_intel 0000:02:00.1: bound 0000:02:00.0 (ops radeon_audio_component_bind_ops [radeon])
> >>
> >> Happy to test further patches or gather more logs.
> >>
> >> Regards,
> >> Erik
>