For newer LSPCon Intel have a patch in DRM-tip that they pointed me to that might be relevant:
I'm going to try applying the patch to the nightly Libreelec kernel this weekend (I have no idea if it will apply ok to that kernel or require use of DRM tip)
If you start to poke around in the Gitlab issue log for the Intel DRM driver you'll see reports and triage that reveal the problem isn't simple like LSPCON vs. no-LSPCON; as there are different LSPCON chips (different chip vendors) with different properties, and even identical LSPCON chips can have rather different implementations (different electrical properties, resulting in different bandwidths available on internal connections) on different boards. In some cases firmware changes can tweak voltages and such that drive the LSPCON chip differently so achieve improvements. In other cases the LSPCON implementation is more fundamentally wrong and no tweaking can get the right result. Most of the issue reports are focused on graphics performance; but the underlying issue is related to bandwidth and thus HBR audio (needing more bandwidth) suffers similarly to higher resolution and refresh rate and colour depth graphics (needing more bandwidth). TL/DR;
Thanks for the tip. I understand your point but the same external adapter and internal LSPCon I list above work fine with Windows for HBR audio which means I know this is a Linux software issue. This is a recurring theme with a lot of these LSPCon issues too; works fine with Windows but broken in Linux. Whilst a sample of one does not mean much the fact the same external DP to HDMI adapter produces some HBR audio with AMD means this might actually be an Intel Graphics driver issue/quirk but I'm hoping for more feedback from others to help confirm or deny this theory.
I realise that this is not a Libreelec specific issue and these problems likely lie in the kernel / firmware but I suspect a high proportion of those that are affected by LSPCON HBR audio issues are Libreelec users so I hope you don't mind me posting this topic here.
Just tested the same external Displayport to HDMI adapter with AMD integrated graphics. The results are different to Intel, HBR audio is intermittent instead of completely silent.
Recently it seems quite a few people have had issue with no HBR Audio when an LSPCON is used. For NUCs this is often a case of update the LSPCon firmware but this is not always the solution / an option.
For anyone reading this and wondering what an LSPCON is or does my setup use one; an LSPCON converts from Displayport to HDMI in this case. Older Intel 7000/8000/9000 CPU motherboards with an HDMI 2.0 port will be using an LSPCON. If you are using an "Active" Displayport to HDMI converter then your setup will also be using an LSPCON. HBR audio is DTS HD Master Audio or Dolby TruedHD with the bitstream option enabled in Kodi (You need to switch to Expert mode and make sure DTSHD and Dolby TrueHD are enabled too).
My experience of this issue is PCM 7.1 and legacy SPDIF Dolby Digital and DTS work fine. I'm interested in hardware and other info from people who have had similar issues of no HBR audio with an LSPCon in the chain.
Below are my findings. Before anyone mentions an immediate firmware update for the LSPCon some of these adapters are known to have firmware which work with HBR audio and they also work fine in Windows. Ultimately I would like to raise an appropriate bug but I need more results in order to spot any pattern (Other than using an LSPCON); I'm not sure if this issue is only when using Intel graphics.
CPU: Intel I3-8300S
Chipset: Intel C246
Video Card: Integrated
LSPCON: Paradetech PS175
LSPCON Type: Integrated on motherboard
LSPCON Firmware Version: 0x2C
General Operation: OK
HDR: OK
HBR Audio: Broken (Silence)
Notes: All OK in Windows
CPU: Intel I3-8300S
Chipset: Intel C246
Video Card: Integrated
LSPCON: Paradetech PS175
LSPCON Type: Integrated on motherboard
LSPCON Firmware Version: 0x40
General Operation: Problematic - No 24/50Hz 4K support
HDR: OK
HBR Audio: Broken (Silence)
Notes: HBR Audio ok in Windows but same 24/50Hz issues as Libreelec.
CPU: Intel I3-8300S
Chipset: Intel C246
Video Card: Integrated
LSPCON: VMM7100
LSPCON Type: External Adapter (CAC-1087)
LSPCON Firmware Version: 7.02.120
General Operation: OK
HDR: OK
HBR Audio: Broken (Silence)
Notes: All OK in Windows. Linux seems to uniquely offer 120hz support with this combination (Assume it is 4K/120/4:2:0 to 4:2:2 conversion).
CPU: AMD 7600X
Chipset: AMD X670E
Video Card: Nvidia RTX3080
LSPCON: VMM7100
LSPCON Type: External Adapter (CAC-1087)
LSPCON Firmware Version: 7.02.120
General Operation: No display after splash screen
Notes: Using required Generic Legacy image.
CPU: AMD 7600X
Chipset: AMD X670E
Video Card: Integrated Graphics
LSPCON: VMM7100
LSPCON Type: External Adapter (CAC-1087) via USB-C passive cable
LSPCON Firmware Version: 7.02.120
General Operation: OK at 4K/60 but not at 4K/24
HDR: Broken (Black screen)
HBR Audio: Broken (Intermittent)
Notes: PCM audio has slight interruptions. SPDIF/legacy bitstream has same interruptions as HBR audio.
sky42 I saw you mentioned the same issue when you tried a Displayport to HDMI adapter; can you provide further info?
I'm not using NUCs but am using a Coffelake/Kabylake motherboard which is using a different brand of Displayport to HDMI converter chip (LSPCON) to your NUCs (Paradetech instead of Megachips).
Still not sure why HBR audio is not working for me but it useful to know the LSPCON/HDMI update fixed it for you.
Changed to a CAC-1088 external LSPCon and this works reliably but still no HBR audio in Linux. Any ideas on how to debug the audio side of HDMI in Linux?
michael123 Can you let me know if they work after the LSPCon update. I have a Coffeelake setup which just will not work with HBR Audio in Linux but is fine in Windows (Seems quite a few people have this issue). I changed from the internal LSPCon to a new Club3D CAC-1088 on the Displayport but although this is consistent/reliable it still does not work with HBR Bitstream in Linux. Some people have suggested that LE10 has less issues with LSPCons but that is not an option for me.
chewitt Thanks for the response. I updated to the LE12 Nightly (19th Jan) and the problem remains the same. I only get basic vesa resolutions alot of the time and dmesg show some issues:
LibreELEC:~ # dmesg | grep i915
[ 0.373362] i915 0000:00:02.0: [drm] VT-d active for gfx access
[ 0.373483] i915 0000:00:02.0: vgaarb: deactivate vga console
[ 0.373523] i915 0000:00:02.0: [drm] Using Transparent Hugepages
[ 0.373529] i915 0000:00:02.0: [drm] *ERROR* Stolen reserved area [mem 0x8f700000-0xbe5fffff] outside stolen memory [mem 0xbc600000-0xbe5fffff]
[ 0.439961] i915 0000:00:02.0: vgaarb: VGA decodes changed: olddecodes=io+mem,decodes=io+mem:owns=io+mem
[ 0.440479] i915 0000:00:02.0: [drm] Finished loading DMC firmware i915/kbl_dmc_ver1_04.bin (v1.4)
[ 1.294324] i915 0000:00:02.0: [drm] [ENCODER:94:DDI A/PHY A] failed to retrieve link info, disabling eDP
[ 1.325912] [drm] Initialized i915 1.6.0 20201103 for 0000:00:02.0 on minor 0
[ 1.400185] fbcon: i915drmfb (fb0) is primary device
[ 1.757923] i915 0000:00:02.0: [drm] fb0: i915drmfb frame buffer device
[ 4.958659] i915 0000:00:02.0: [drm] *ERROR* Link Xraining UnsuccessfulThe EDID is for the correct device. Only the first displayport shows as connected which I believe is correct for LSPcon. The errors and end effect is the same if I use a VM or bare metal.
i915_display_info is here:
cat /sys/kernel/debug/dri/0/i915_display_infoCRTC info---------[CRTC:51:pi - Pastebin.com
The only thing that looks odd to me is it reporting HDCP 1.4 when the whole chain should support HDCP 2.X (Other than the completely wrong resolution table).
Are there any kernel switches I can use to increase the logging from i915?
I've now tried on bare metal as well as with my VM and the behaviour is identical which means I can rule out the VM as being the cause. From what I have read most of these sorts of issues were solved 3-5 years ago in the kernel so I'm not sure if this is a general Linux/hardware issue or specific to Librelec.
Here is my dmesg from the VM:
[ 0.000000] Linux version 6.1.68 (sky42@jammy) (x86_64-libreelec-linux-gnu-gc - Pastebin.com
The issue is the same whether I use official 11.0.4 or the Sky42 build.
The Intel NUCs which use a Paradetech PS175 LSPCon seems to have a major firmware version of 7 (eg. 0x07_0x020) but my board seems to be on 0x08_0x2C.
Is there any more info I can provide to try to get some support with this?
The HDMI 2.0 output on my motherboard uses a Paradetech LSPCon. In Windows bitstream audio works fine with Kodi. With the Libreelec 11.0.4 though I get no sound at all with bitstream audio (PCM up to 7.1 works). Any tips on config changes I could try? (I am selecting the same working HDMI device for bitstream that I use for PCM).
On the plus side HDR switching is working nicely.
Display MoreHmm strange yes in the Datasheet it is clearly mentioned... But when you pull up the data sheet of the CPU you can see that over HDMI it only supports 24Hz and not 60hz and the NUC10i3FNHN1 has only one HDMI port: (https://ark.intel.com/content/www/us…o-4-10-ghz.html)
I installed both the LE and nightly build. With the LE version I have no sound over HDMI, it can't find any Audio output devices.
With the nightly version (20220420) I get sound however only 2 ch and it does not deliver ATMOS or anything advanced to my speaker system (even tough my media file do contains such audio streams).
I don't know why I cant pass through the sound to the speaker system since I know when I hook up my PC i can easily pass ATMOS etc. through my HDMI but his PC has an dedicated GPU.
Andy
I suspect it is using an LSPCON chip. This converts the Display port output from the CPU to HDMI 2.0 and allows for 4K/60 over HDMI.
Hi All,
I'm using Arch Linux with KVM VMs. The modern and officially supported by Intel way to passthrough a whole graphics card with it's HDMI display output is to use GVT-D with UPT. My host works when using this with a Windows VM so I know the host side / hardware is good.
The setup uses an emulated display for the bios boot and then passes through the Intel integrated GPU as a secondary display. No need for option rom or option rom mapping.
When I try to use this with the Generic Libreelec image it starts to boot and the white mouse pointer appears on the Intel GPU HDMI display output but it then hard reboots. Is there any appetite to get this working with Libreelec?
Here is the relevant bits from a virsh VM config file:
<devices>
<graphics type='vnc' port='5901' autoport='no' listen='0.0.0.0'/>
<hostdev mode='subsystem' type='pci' managed='yes'>
<driver name='vfio'/>
<source>
<address domain='0x0000' bus='0x00' slot='0x02' function='0x0'/>
</source>
<alias name='hostdev0'/>
<address type='pci' domain='0x0000' bus='0x00' slot='0x03' function='0x0'/>
</hostdev>
</devices>
<qemu:commandline>
<qemu:arg value='-set'/>
<qemu:arg value='device.hostdev0.x-igd-opregion=on'/>
<qemu:arg value='-set'/>
<qemu:arg value='device.hostdev0.x-igd-gms=1'/>
</qemu:commandline>Tried a PR but it seems only VirtIO is wanted and no love for SRIOV but thanks anyway.
Krobar submit a pull request on github with a reasonable explanation for why the config change is needed and it will be considered. As a general rule enabling a couple of minor drivers isn't a big deal and shouldn't be an problem.
Have submitted a pull request, its my first public PR so please excuse me if it isn't quite right.
You need to adjust linux kernel config LibreELEC.tv/linux.x86_64.conf at master · LibreELEC/LibreELEC.tv · GitHub. Both
CONFIG_IGBVF and CONFIG_IXGBEVF are not set. Just change both to y
CodeCONFIG_IGBVF=y CONFIG_IXGBEVF=y
Thanks, I know 
It works and I have been running from my last build for months. What I wanted was try to get it included as standard in the normal Libreelec builds.
I don't know why people cause complication for themselves with obscure formats. Go read the advice in post #2 again.
Although raw DSD is difficult to work with form a DSP point of view it is a different format to PCM and all conversion to PCM are in some way lossy. So the original poster is correct to store DSD in its native format from an archive/quality POV.
Thanks again Vpeter, there was an old hosts file entry which I must have put there months ago.
Out of interest what would I need to do to get the Intel igbvf and ixgbevf Virtual Functions included in the mainline? (They are just a kernel option and the only reason I have to build from source)
The DSD tracks play fine on an Oppo BDP-103. I'm not having much luck with DSD on an Odroid C2, either. The tracks are 5.1 and I suspect that the transcoding is just too much to sustain at that bitrate.
I'm probably something of an outlier in that I appreciate the video capabilities of LibreELEC, but my primary interest is streaming multichannel audio from DVD-A and SACD rips. To that end, I'm finding that the C2 is definitely a much better choice than even the newest Pi. While my original post said I was no longer getting dropouts on the Pi, that unfortunately didn't last. The Odroid seems quite solid, though.
Really could do with native DSD output from HDMI but not sure who to ask about this; is Libre using manufacturer drivers for video cards? (Guessing Noveau isn't yet stable enough). FFMPEG and the rest of the Linux subsystem can detect and work with DSD but there is not yet any support from Nvidia/AMD/Intel.
