VDI Options | Post VMware Roadmap / Decisions

[dcsapak]

the latest commit (as of this writing) in backport/main does not build with kernel 6.8.12-1-pve for me.
the same it seems to build OK and work fine with kernel 6.8.8-3-pve

intel flex 140 with FW: DG02_2.2356

thanks, i could reproduce it and opened an issue on the upstream git repo: https://github.com/intel-gpu/intel-gpu-i915-backports/issues/190

 

[lleo_]

so Intel acknowledged the issue we reported, and had two additional version commits since then.
For me, these still do not build, with the same error as reported back in August.
Intel verified and claims it is fixed and points to a 'custom' kernel.
Just to mention it again, the latest commit does build and work OK on 6.8.8-4-pve, but not on 6.8.12-4-pve

You previously stated to them the PVE kernels are based on Ubuntu server versions with some patches added on.
Could it be that some PVE patch is causing this issue?

Thanks for looking into this

Separately: any progress on the guide for using Intel DataCenter GPUs in PVE?

 

[shabsta]

I’m also waiting for an update

 

[dcsapak]

Hi,

sorry for taking so long, have not had time for this recently.

I checked the state and they fixed it indeed for newer ubuntu kernels, but their ubuntu version detection don't trigger with our kernel (which makes sense). I asked them about possible solutions here:
https://github.com/intel-gpu/intel-gpu-i915-backports/issues/203

Let's see what they answer, but in the meantime you'd have to patch their repo manually with the following diff to make it work (note that these changes will only work on our/ubuntus kernel, and not an upstream 6.8)

diff --git a/backport-include/backport/backport_macro.h b/backport-include/backport/backport_macro.h
index 27a2019..55ba361 100644
--- a/backport-include/backport/backport_macro.h
+++ b/backport-include/backport/backport_macro.h
@@ -5,7 +5,7 @@
 #include <backport/autoconf.h>
 
 #if LINUX_VERSION_IS_GEQ(6,9,6) || \
-       (LINUX_VERSION_IS_GEQ(6,8,0) && UBUNTU_BACKPORT_VERSION_IS_GEQ(43,43))
+       LINUX_VERSION_IS_GEQ(6,8,0)
 
 /*
  * 8e21de5f99b2 drm/mst: Fix NULL pointer dereference at drm_dp_add_payload_part2

If there isn't a good way forward with their repo, I could fork it with these changes and one could use that...

 

[lleo_]

Hi,

sorry for taking so long, have not had time for this recently.

I checked the state and they fixed it indeed for newer ubuntu kernels, but their ubuntu version detection don't trigger with our kernel (which makes sense). I asked them about possible solutions here:
https://github.com/intel-gpu/intel-gpu-i915-backports/issues/203

Let's see what they answer, but in the meantime you'd have to patch their repo manually with the following diff to make it work (note that these changes will only work on our/ubuntus kernel, and not an upstream 6.8)

diff --git a/backport-include/backport/backport_macro.h b/backport-include/backport/backport_macro.h
index 27a2019..55ba361 100644
--- a/backport-include/backport/backport_macro.h
+++ b/backport-include/backport/backport_macro.h
@@ -5,7 +5,7 @@
 #include <backport/autoconf.h>
 
 #if LINUX_VERSION_IS_GEQ(6,9,6) || \
-       (LINUX_VERSION_IS_GEQ(6,8,0) && UBUNTU_BACKPORT_VERSION_IS_GEQ(43,43))
+       LINUX_VERSION_IS_GEQ(6,8,0)
 
 /*
  * 8e21de5f99b2 drm/mst: Fix NULL pointer dereference at drm_dp_add_payload_part2

If there isn't a good way forward with their repo, I could fork it with these changes and one could use that...