thanks ! That’s what I was looking for 
There seems to be a regression between RC3 & RC4 regarding GPU passthrough, at least for me.
I was able to passthrough the GPU using the stubroot method as well as xml method with RC3 (although I did get a strange error where non-HVM qubes would fail to start with an ‘Memory Hot Plug’ error with the xml method). I don’t have a way to turn off ReBAR so I had to change parameter 3.5G to 2G with both methods in order to get it to work. When I installed RC4, however, I was unable to passthrough the gpu with either methods.
Could anyone tell me if there was a change in Xen/Kernel/Qubes that would cause this to happen?
And which Xen/kernel/Qubes do you use? (Guessing RC means Qubes v4.2 RCx might make some people not to respond considering it lazy writing.)
sure that makes sense. You can’t expect much help if you don’t provide enough information for people to help.
Yes, by RCx I meant ‘release candidate’. I’ve installed each Qubes 4.2 release candidate with kernel latest, so with RC3 I installed it with kernel 6.4.8-1 and with RC4 I installed it with kernel 6.5.x-x (can’t recall it by memory).
Aside from that, I did not make any modifications except for the ones necessary for GPU passthrough
What’s the best place to report that issue?
(Should be in a dedicated “user support” thread)
What is the issue/what is the thing that don’t work ? (related to tolud ? something else ? )
What GPU are you using ?
What GPU driver are you using ?
What OS are you using for the HVM ?
Do you have the qubes logs of this HVM ?
Do you have the result of “sudo dmesg” for this HVM ?
In case anyone runs into this issue, the following worked:
But this topic is about “... lots of RAM”? And your post is about 2GB RAM, or I misread it?
I’m hoping I’m not causing a lot of confusion about the topic since I’m posting in a couple of threads.
When I refer to 2GB, I’m referring to the parameter in the stubroot method, where, in my case, I have to use the 2GB parameter instead of the usual 3.5GB parameter in order to get the gpu passed-through to a qube with lots of ram.
In my case, after upgrading dom0 from 4.2 rc3 to 4.2 rc4 or 4.2 rc5, the passthrough qube would just crash, but after following the gpu pass-through guide after updating dom0, the qube worked again.
Just wanted to post this here, in case any one else runs into the issue.
1 Like