On 30/09/26 09:57 PM, Bernhard Schmidt wrote: > After upgrading to todays security kernel a pair of BCM57416 onboard > ports fail after a couple of minutes uptime with the following error > message in the log > > Sep 30 21:33:27 nm3 kernel: Linux version 6.12.111+deb13-amd64 > ([email protected]) (x86_64-linux-gnu-gcc-14 (Debian 14.2.0-19) > 14.2.0, > [...] > Sep 30 21:33:27 nm3 kernel: bnxt_en 0000:19:00.0 eth0: Broadcom BCM57416 > NetXtreme-E 10GBase-T Ethernet found at mem 9dc10000, node addr > 00:0a:f7:e7:f0:5c > Sep 30 21:33:27 nm3 kernel: bnxt_en 0000:19:00.0: 63.008 Gb/s available PCIe > bandwidth (8.0 GT/s PCIe x8 link) > Sep 30 21:33:27 nm3 kernel: bnxt_en 0000:19:00.1 (unnamed net_device) > (uninitialized): Device requests max timeout of 100 seconds, may trigger hung > task watchdog (kernel default 120s) > Sep 30 21:33:27 nm3 kernel: bnxt_en 0000:19:00.1 eth1: Broadcom BCM57416 > NetXtreme-E 10GBase-T Ethernet found at mem 9dc00000, node addr > 00:0a:f7:e7:f0:5d > Sep 30 21:33:27 nm3 kernel: bnxt_en 0000:19:00.1: 63.008 Gb/s available PCIe > bandwidth (8.0 GT/s PCIe x8 link) > Sep 30 21:33:27 nm3 kernel: bnxt_en 0000:19:00.0 eno1np0: renamed from eth0 > Sep 30 21:33:27 nm3 kernel: bnxt_en 0000:19:00.1 eno2np1: renamed from eth1 > [...] > Sep 30 21:37:42 nm3 kernel: DMAR: DRHD: handling fault status reg 2 > Sep 30 21:37:42 nm3 kernel: DMAR: [DMA Read NO_PASID] Request device > [19:00.0] fault addr 0xefffc000 [fault reason 0x06] PTE Read access is not set
A friend reminded me that they were seeing the same thing after a Proxmox kernel upgrade from 6.7.13-21-pve to 7.0.14-12 (so across major versions). It was reported to Proxmox and a fix is not known. However, as workaround you can set the IOMMU in Passthrough mode (iommu=pt on the kernel commandline) which fixes the problem. This also works for 6.12.111+deb13-amd64 Bernhard
