HDR turns off and back on every few seconds on NUC6

Whenever I enable HDR on my NUC6, the output frequently drops out of HDR10 and then immediately switches back into HDR10. I can see this because the TV displays the indicator HDR10 to show it is receiving an HDR stream, then after a short while the image freezes and dims very briefly before resuming and the TV displays the HDR10 indicator again. This usually occurs every few seconds but sometimes it appears to happen a few times per second and other times there might be a pause of ten seconds or so before it happens.

All the following scenarios trigger the issue, which typically occurs within a second or two of HDR becoming enabled:

• Kodi v21 (compiled for GLES) in GBM mode playing an HDR stream running with any of the kernels drm-tip, 6.8.0-36.36 (Ubuntu stock), 6.9.3 (Ubuntu mainline)

• Ubuntu 24.04 gnome on kernel 6.8.0-36.36 with Wayland experimental HDR enabled (ie with global.compositor.backend.get_monitor_manager().experimental_hdr = 'on' entered in looking glass).

• Neon Plasma6 with kernel 6.5 and HDR turned on (via the Display settings and with /etc/environment edited to include KWIN_DRM_ALLOW_INTEL_COLORSPACE=1)

To eliminate issues with the cables or TV, I have tried with 2 different HDMI cables and confirmed that the issue does not occur when using these cables connected to a desktop with an NVIDIA 4070 card running Windows with HDR enabled.

With kernels 6.15 and 6.18, I see an error that looks related because it looks like i915 was expecting SDR but saw HDR. It doesn't happen in kernels 6.9.3 and drm-tip, though. This example is from the neon kernel, but 6.8 gives an identical error):

Jul 08 03:28:51 neon kernel: i915 0000:00:02.0: [drm] ERROR [CRTC:59:pipe A] mismatch in infoframes.enable (expected 0x00000006, found 0x00000002)

Jul 08 03:28:51 neon kernel: i915 0000:00:02.0: [drm] ERROR mismatch in vsc dp sdp

Jul 08 03:28:51 neon kernel: i915 0000:00:02.0: [drm] ERROR expected:

Jul 08 03:28:51 neon kernel: i915 0000:00:02.0: DP SDP: VSC, revision 5, length 19

Jul 08 03:28:51 neon kernel: i915 0000:00:02.0: pixelformat: RGB

Jul 08 03:28:51 neon kernel: i915 0000:00:02.0: colorimetry: BT.2020 RGB

Jul 08 03:28:51 neon kernel: i915 0000:00:02.0: bpc: 8

Jul 08 03:28:51 neon kernel: i915 0000:00:02.0: dynamic range: CTA range

Jul 08 03:28:51 neon kernel: i915 0000:00:02.0: content type: Not defined

Jul 08 03:28:51 neon kernel: i915 0000:00:02.0: [drm] ERROR found:

Jul 08 03:28:51 neon kernel: i915 0000:00:02.0: DP SDP: VSC, revision 0, length 0

Jul 08 03:28:51 neon kernel: i915 0000:00:02.0: pixelformat: RGB

Jul 08 03:28:51 neon kernel: i915 0000:00:02.0: colorimetry: sRGB

Jul 08 03:28:51 neon kernel: i915 0000:00:02.0: bpc: 0

Jul 08 03:28:51 neon kernel: i915 0000:00:02.0: dynamic range: VESA range

Jul 08 03:28:51 neon kernel: i915 0000:00:02.0: content type: Not defined

Since the NUC6 uses a DP-HDMI converter internally, I tried both with and without enable_dp_mst=0 in case this made any difference, but it didn't.

Other info:

uname -a Linux NUC 6.10.0-rc6+ #1 (moved) SMP PREEMPT_DYNAMIC Mon Jul 8 20:07:39 AWST 2024 x86_64 x86_64 x86_64 GNU/Linux

lspci -vnn -d :*:0300 00:02.0 VGA compatible controller [0300]: Intel Corporation HD Graphics 500 [8086:5a85] (rev 0b) (prog-if 00 [VGA controller]) DeviceName: CPU Subsystem: Intel Corporation HD Graphics 500 [8086:2067] Flags: bus master, fast devsel, latency 0, IRQ 135, IOMMU group 0 Memory at 90000000 (64-bit, non-prefetchable) [size=16M] Memory at 80000000 (64-bit, prefetchable) [size=256M] I/O ports at f000 [size=64] Expansion ROM at 000c0000 [virtual] [disabled] [size=128K] Capabilities: [40] Vendor Specific Information: Len=0c <?> Capabilities: [70] Express Root Complex Integrated Endpoint, MSI 00 Capabilities: [ac] MSI: Enable+ Count=1/1 Maskable- 64bit- Capabilities: [d0] Power Management version 2 Capabilities: [100] Process Address Space ID (PASID) Capabilities: [200] Address Translation Service (ATS) Capabilities: [300] Page Request Interface (PRI) Kernel driver in use: i915 Kernel modules: i915

Machine: NUC6CAYH Vendor: Intel Corp. Version: AYAPLCEL.86A.0076.2023.0710.1018 Release Date: 07/10/2023 HDMI (lspcon) firmware: 1.60

changed the description

I've got a drm-tip log with debug enabled, but I don't see the "paperclip attach icon" anywhere on this page to upload it. Do I need to have more permissions for this?

See the comment box toolbar for the attach icon.

While I figure out how to upload the log you can see it at https://drive.google.com/file/d/1YPYYZ0VjOpG6JnGUfHmV6SaRf3xujipQ/view

And the vbt file is here: https://drive.google.com/file/d/10u-e5aLuph12weMH1VnAMdgOBaZpl4Ne/view

changed the description

added Community feature: display/color labels

Cc : @uma_mnnit @ckborah

How often this bug occurs seems to be related to the display refresh rate. It's at least every few seconds at 60 Hz, a bit less frequent at 50 Hz, and at 30 Hz it might take a minute or even several minutes before it occurs.

What's with the EDID?

i915 0000:00:02.0: [drm:drm_edid_load_firmware] [CONNECTOR:115:DP-1] Loaded external firmware EDID "tcl-c855.edid"

I experienced an issue whereby the NUC would lose the resolution lines after the TV was put into standby and default to some ridiculously lowres setting until it was rebooted, so I dumped the edid into a file and loaded it via the kernel boot args. It might not be necessary because I also had to specify the the resolution to get it to work properly. GRUB_CMDLINE_LINUX_DEFAULT is:

quiet splash scsi_mod.use_blk_mq=1 drm.edid_firmware=DP-1:tcl-c855.edid video=DP-1:3840x2160@60 enable_dp_mst=0

Okay. May be irrelevant, but please let's rule that out and remove drm.edid_firmware=DP-1:tcl-c855.edid video=DP-1:3840x2160@60 enable_dp_mst=0 from the params during debugging.

Ok, will do. I'm having trouble doing a git pull, though - it times out or says connection refused. Trying a new clone fails, too. This is as far as it gets before saying it timed out:

git clone git://anongit.freedesktop.org/drm-tip -v
Cloning into 'drm-tip'...
Looking up anongit.freedesktop.org ... done.
Connecting to anongit.freedesktop.org (port 9418) ... 131.252.210.161 done.

Has the location of the repo changed or is there an issue with it, or am I just doing something wrong?

Please try with https://gitlab.freedesktop.org/drm/tip.git

I've done a new logfile with drm.debug=0xe and the drm.edid_firmware, video, and enable_dp_mst kernel args removed and running with the latest drm-tip. It's at https://drive.google.com/file/d/13MJWV8wfhHNYZa7U7dx0LmMOY1vSyyhD/view