On Sat, 1 Nov 2025 at 09:43, Richard Henderson <[email protected]> wrote: > > On 10/31/25 21:21, Michael Tokarev wrote: > >> Suggested-by: Peter Maydell <[email protected]> > >> Resolves: https://gitlab.com/qemu-project/qemu/-/issues/3171 > >> Signed-off-by: Richard Henderson <[email protected]> > > > > I wonder if we should pick this up for qemu-stable too (including > > 10.0.x lts series). It's a rather large change though. > > > > The patch applies cleanly to both 10.0.x and 10.1.x series, and > > seems to be working fine. Maybe it can be picked up for a later > > release. > > I think you should hold off for now. > > > What's the implication of this bug, anyway? > > Failure to raise an alignment trap accessing Device memory when > architecturally required > with -smp 1.
We have failures both ways, I think -- the bug report is about an alignment trap we raise when we should not, for a 4-aligned LDRD to Device memory. There are probably also cases where we don't trap but we should. But I think this is something of a corner case, so I agree with postponing it for a bit, so we can see if it has any unexpected regressions in trunk. thanks -- PMM
