On Tue, 07 May 2013 22:15:07 +0200 Christoph Egger <christoph@debian.org> wrote: > FWIW I've been seeing this with gcc-4.6 gcc-4.7 and clang -- that's > the reason we still don't have a newer kfreebsd-10 upload in the > archive. Thank you for starting to dig! This is also the reason we > have no 9.1 in kfreebsd-9 at least in experimental -- all tries didn't > boot. All on my Notebook (Thinkpad X220 -- Sandy Bridge I5) Thanks, reassuring to know it's not just me. =) I've just spent a while looking over some strace output from the KVM guest process when this happens, and I don't see anything that looks concerning. top indicates that the first child thread of the qemu-kvm process (presumably the worker for vCPU 0) is using 100% CPU time, but strace says it's in the middle of an ioctl call: [pid 28279] ioctl(11, 0xae80 <unfinished ...> lsof tells me that fd 11 in this case is of type anon_inode. As far as I can tell, this is normal for a KVM guest process. I tried looking at a healthy guest on the same machine, and it's behaving in the same way -- it looks like this ioctl is the way that KVM vCPUs wait for something to do. I think this is ultimately not an efficient means of investigation to pursue, so I plan to drop it and go back to trial and error -- I'll add more and more of the responsible function definition until it breaks, and then we can point at a specific line of code. I probably won't have time to do this until the weekend, but I'll post back here when I have something more to add.
Attachment:
signature.asc
Description: PGP signature