There is an issue in the kernels 6.1.130 & 6.1.131, while in 6.13.6 & 6.13.7 it works ok. Cannot start VM with error 2025-03-08T14:53:10.620302Z qemu-kvm: VFIO_MAP_DMA failed: Cannot allocate memory 2025-03-08T14:53:10.620323Z qemu-kvm: vfio_container_dma_map(0x5621ca6e5400, 0xc0000, 0x20000, 0x7fd550600000) = -12 (Cannot allocate memory) qemu: hardware error: vfio: DMA mapping failed, unable to continue Config & logs are in the attachment.
Created attachment 307842 [details] grub.cfg
Created attachment 307843 [details] VM config
Created attachment 307844 [details] Successful start on 6.1.129
Created attachment 307845 [details] Error on 6.1.130
Created attachment 307846 [details] Error on 6.1.131
Created attachment 307847 [details] Ok in 6.13.6
Created attachment 307848 [details] Ok in 6.13.7
There is the same bug in 6.1.132 and it's ok in 6.14.0.
Created attachment 307905 [details] 6.1.132
Created attachment 307906 [details] 6.14.0
Just wanted to add a note that elrepo does not make modifications to the source code from kernel.org. Chances are this is a bug in linux-6.1.130. The mainline kernels do not show any problem. Assuming that some backport caused the issue, there might be additional patches that should have been applied together.
Same, started today after upgrading Debian kernel to 6.1.0-33-amd64.
*fix 6.1.133-1
Ok in 6.1.129-1 (previous Debian kernel) Error in 6.1.133-1