[SOLVED] PCI passthrough onboard sound

[diversity]

thx all for the help. finally managed to get the sound passed through. I am happy

 

[koivum]

thx all for the help. finally managed to get the sound passed through. I am happy

Awesome, glad you got it all to work. Enjoy

 

[HDN]

Hi

I have faithfully followed the steps you propose, but I have problems at the time of compilation.

The last output in the shell are:
....
dpkg-buildpackage: info: full upload (original source is included)
lintian zfs-linux_0.8.5-pve1.dsc
W: zfs-linux source: debian-rules-uses-unnecessary-dh-argument dh ... --parallel (line 13)
W: zfs-linux source: package-file-is-executable debian/changelog
W: zfs-linux source: package-file-is-executable debian/control
W: zfs-linux source: package-file-is-executable debian/copyright
E: zfs-linux source: executable-debhelper-file-without-being-executable debian/libnvpair1linux.docs
E: zfs-linux source: executable-debhelper-file-without-being-executable debian/libnvpair1linux.install
E: zfs-linux source: executable-debhelper-file-without-being-executable debian/libuutil1linux.docs
E: zfs-linux source: executable-debhelper-file-without-being-executable ... use --no-tag-display-limit to see all (or pipe to a file/program)
W: zfs-linux source: spelling-error-in-patch-description debian/patches/0006-dont-symlink-zed-scripts.patch dont don't
W: zfs-linux source: newer-standards-version 4.5.0 (current is 4.3.0)
make[1]: *** [Makefile:54: zfs-linux_0.8.5-pve1.dsc] Error 1
make[1]: Leaving directory '/usr/src/pve-kernel/build/modules/tmp'
make: *** [Makefile:99: pkg-zfs.prepared] Error 2


Do you know how to solve it?
Thanks

 

[koivum]

Hi

I have faithfully followed the steps you propose, but I have problems at the time of compilation.

The last output in the shell are:
....
dpkg-buildpackage: info: full upload (original source is included)
lintian zfs-linux_0.8.5-pve1.dsc
W: zfs-linux source: debian-rules-uses-unnecessary-dh-argument dh ... --parallel (line 13)
W: zfs-linux source: package-file-is-executable debian/changelog
W: zfs-linux source: package-file-is-executable debian/control
W: zfs-linux source: package-file-is-executable debian/copyright
E: zfs-linux source: executable-debhelper-file-without-being-executable debian/libnvpair1linux.docs
E: zfs-linux source: executable-debhelper-file-without-being-executable debian/libnvpair1linux.install
E: zfs-linux source: executable-debhelper-file-without-being-executable debian/libuutil1linux.docs
E: zfs-linux source: executable-debhelper-file-without-being-executable ... use --no-tag-display-limit to see all (or pipe to a file/program)
W: zfs-linux source: spelling-error-in-patch-description debian/patches/0006-dont-symlink-zed-scripts.patch dont don't
W: zfs-linux source: newer-standards-version 4.5.0 (current is 4.3.0)
make[1]: *** [Makefile:54: zfs-linux_0.8.5-pve1.dsc] Error 1
make[1]: Leaving directory '/usr/src/pve-kernel/build/modules/tmp'
make: *** [Makefile:99: pkg-zfs.prepared] Error 2


Do you know how to solve it?
Thanks

Hi,
Which command did you run for make? Did you run:

make 2>&1 | tee build.log

If so, could you check file build.log and look for errors there? They tend to be more explicit about the problem than shell output.
Thanks

 

[HDN]

Hi,
Which command did you run for make? Did you run:

make 2>&1 | tee build.log

If so, could you check file build.log and look for errors there? They tend to be more explicit about the problem than shell output.
Thanks

Hi

I attached the build.log file.

Thanks for your help.

Regards

 

[randomman87]

Does anyone know if this was patched in the repositories? I suspect I'm having the same issue only with the AMD Matisse USB controller, the sound card passed through fine. Whole system becomes unstable when I start a VM with the USB controller, syslog gives the same FLR errors.

I've passed through USB devices, but am having USB lag issues, and when I use my USB switch (KVM without video) there's a larger delay than native Windows. PCIe passthrough may not solve it but I want to try.

 

[diversity]

I do not believe it has been fixed as I tried passing the USB controller (as PCI(e) pass trough) some time ago can't remember which exact pve 7 version) and things started to FLR on me again.
I then reverted back to only passing through the USB audio devices (as USB) and things went back to working for me in terms of audio.

What still does not work for me though is passing through an android phone (as USB) as windows server 2019 will not recognize the device and even installing samsing USB drivers will not solve the issue.

All in all I think USB passthrough (either the controller as PCI(e) or as USB device)is still a hit and miss and does not get the love it needs.

 

[randomman87]

Have you tried patching the kernel as mentioned in this thread? Or is that different now that we're on PVE 7?

What does lsusb show for the Android phone on the host?

Absolutely, PCIe passthrough too.

 

[diversity]

i did do the kernel patch with help of this thread on pve 6 and had joy for a while. but as soon as i realized i actually did not need that as passing through the usb audio device as usb did the trick for me as well I moved to that solution directly after updating to pve 7.

i am too inexperienced to comfortably patch stuff and not gladly do that again any time soon if i can avoid it.

regarding lsusb on host it shows;.
Bus 011 Device 016: ID 04e8:6860 Samsung Electronics Co., Ltd Galaxy A5 (MTP)
In the pve gui (vm >: hardware) it shows as;
SAMSUNG_Android

Now the thing is it is a samsung galaxy note 9 that is attached and not a galaxy A5.
Do you have advice or should I make a new thread?

EDIT; a plantronics headset USB passedthrough does register normally on windows server 2019 making use of the same usb port (11-1, as 11-1 is the usb port that is passed through, not a specific device although passing through the samsung android usb device specifically also does not work)

 

[diversity]

@randomman87 this morning I received a message from logwatch;
WARNING: Kernel Errors Present
usb 11-1: device descriptor read/64, error -71 ...: 3 Time(s)

could this be related to my android not being recognized by the vm?

 

[randomman87]

I'm not the best one to ask, I'm pretty much a Linux noob but I believe 11-1 means bus 11 device 1 and your previous post has the Android as device 16

I found more information about the FLR quirk. The USB and Audio FLR quirks should now be in PVE kernel as they're in the Ubuntu master quirks.c, and PVE git says they use Ubuntu kernel. However, there's a new known issue with the Encryption Controller that is part of the same "device" as the USB and audio, which causes VMs with those devices passed through not to boot.

Does PVE need TPM? If not I suspect if I disable fTPM in the BIOS that this device will disappear and no longer be a problem.

EDIT: Nope, nothing to do with fTPM. It's apparently a hardware accelerated cryptoprocessor (as the name should have implied) and is used for crypto related tasks like hashes etc. Doesn't look to be a way to disable it in UEFI. Reading the patch thread I linked to above it looks like talk stalled on the 29th of September 2021 with AMD (I think) suggesting that a FLR quirk shouldn't be added as it could be an attack vector if the device is re-used on another guest without a proper reset, and basically saying they don't care because it's a consumer device.

 

[eminempt]

how confusing this was.. for me it was only a matter of adding the corresponding PCI device on the PxMx GUI

 

[joysthokkins]

Hi! Thanx a lot!

I have this error:

applying patch '../../patches/kernel/0008-amd-no-flr.patch'
patching file drivers/pci/quirks.c
Hunk #1 FAILED at 5040.
1 out of 1 hunk FAILED -- saving rejects to file drivers/pci/quirks.c.rej
make: *** [Makefile:108: ubuntu-kernel.prepared] Error 1