On Sun, Feb 22, 2026 at 8:42 PM Helmut Grohne <[email protected]> wrote:
> Hi, > > On Sun, Feb 22, 2026 at 04:47:22PM +0100, Paul Gevers wrote: > > Please be aware that the autopkgtest of debvm is doing the same. From the > > log [1]: > > Thank you for highlighting this to me. As far as I understand things, > the critical piece that makes things stuck is LPA2 (52bit address > support). I tried disabling this. What I get is: > > https://salsa.debian.org/helmutg/debvm/-/jobs/9095840 > | UEFI firmware (version 2025.11-4 built at 01:31:14 on Feb 17 2026) > | Synchronous Exception at 0x47EFE008 > > Before this change, I only got: > > https://salsa.debian.org/helmutg/debvm/-/jobs/9076802 > | UEFI firmware (version 2025.11-4 built at 01:31:14 on Feb 17 2026) > > So this seems to indeed make things proceed further, but only so much. > Any further ideas? > > Hello It is not enough to disable LPA2 The workaround I proposed is to switch to another CPU model: neoverse-n1 I don't think that is a hard requirement to use -cpu max > Completely unrelated, I also got: > > https://salsa.debian.org/helmutg/debvm/-/jobs/9095842 > | [ 22.935909] ------------[ cut here ]------------ > | 1664 > | [ 22.936407] kernel BUG at arch/x86/kernel/alternative.c:2516! > | 1665 > | [ 22.939517] Oops: invalid opcode: 0000 [#1] SMP NOPTI > | 1666 > | [ 22.940644] CPU: 0 UID: 0 PID: 311 Comm: (udev-worker) Not tainted > 6.18.12+deb14-amd64 #1 PREEMPT(lazy) Debian 6.18.12-1 > | 1667 > | [ 22.941476] Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS > 2025.11-4 02/17/2026 > > This one is not reproducible (neither on salsa nor locally). > > I'm a bit stuck on all these hangs, but qemu seems to be stuck migrating > to forky due to this. > > Helmut > > -- Hector CAO Software Engineer – Server Team / Virtualization [email protected] https://launc <https://launchpad.net/~hectorcao>hpad.net/~hectorcao <https://launchpad.net/~hectorcao> <https://launchpad.net/~hectorcao>
