Eric Augé <[email protected]> writes: > Hello, > > small questions regarding 2 things I noticed today: > > II == vmd guest: linux x86 host: openbsd/amd64 == > > I had a running/installed x86 alpine linux virt VM running on vmm(4). > I "sysupgrade -s" 'd 2 days ago from my previous -current and the x86 > guest just does not run anymore, I've tried reinstalling an > alpine-virt 3.22 x86, vmd goes to 100% CPU after a minute or 2 when > booting the virt install ISO. > > https://dl-cdn.alpinelinux.org/alpine/v3.22/releases/x86/alpine-virt-3.22.1-x86.iso > > I had not run it for a while, so I am not sure when the behavior changed. > I am not sure if it's a me problem, or if anybody else experiences > similar symptoms on linux x86 guests on an amd64 OpenBSD host HV.
Can you share dmesg output of the host? Specifically cpu0 details. If you boot a linux guest and can toggle its boot args, add console=ttyS0,115200 and remove "quiet" if its present. Ideally you'd be able to share the output of the Linux kernel messages prior to any hang. There were changes to vmm that would affect Intel-based hosts but without the cpu0 details I can't say if your machine would be impacted. > > I tried with OpenBSD i386 and it seems to run fine > > II == vmctl reload == > > I noticed another thing, "vmctl reload" just stalls... and never ends > it seems to stall on recvmsg() from /var/run/vmd.sock > --- > 6808 vmctl RET kbind 0 > 6808 vmctl CALL kbind(0x734d85a9f0e0,24,0xd4092a5288fecfd1) > 6808 vmctl RET kbind 0 > 6808 vmctl CALL recvmsg(3,0x734d85a9f1d0,0) > ... > --- > Anyone experienced something similar ? > Any idea what I have missed? > I see this on my machine. I'll look at it. Probably a bug in some recent imsg changes I committed in the past month or so. Thanks for reporting. > Thank you, > Best regards, > Eric. > > [2. application/octet-stream; dmesg.boot]...
