Hi Frederik, On Tue, Nov 03, 2026 at 06:17:26PM +0100, Frederik Himpe wrote: > Package: src:linux > Version: 6.12.115-2 > Severity: normal > X-Debbugs-Cc: [email protected], [email protected] > User: [email protected] > Usertags: amd64 > > I updated from linux-image-amd64 version 6.12.113-1 to 6.12.115-2 > yesterday evening. Today as soon as users started using the system with > some load, the network completely becomes unresponsive. tcpdump only > shows outgoing ARP responses and nothing else. When booting the system > with iommu=pt, the netwok continues to work even with load. > > This sounds a lot like bug #1149564 > https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=1149564 > but this one was supposed to be fixed in 6.12.113. Has this regressed > again? > > I attached a kernel log of a session without iommu=pt where it just hangs. > > I have not seen this problem on this system with > 6.12.111, nor 6.12.113 on this system (Dell PowerEdge R7425, AMD CPUs), > although it might theoretically be possible that not enough load was > caused to trigger this issue. It seems unlikely though, seeing how it > now happens after a few minutes.
Are you able to reasily reproduce the problem and can spare an affected machine for experiments? In that case I would like to ask if you could test again 6.12.113-1 based kernel to have confidence it is not broken there, and reconfirm then on 6.12.115-1. As a first step though check if 6.12.117-1 currently in trixie still triggers the problem. If yes then I would suggest we try to bisect the change to have something at hand to report upstream. Worst case we can as well report as it is. Regards, Salvatore

