Hi Milan, On 8/5/20 8:53 AM, Milan Kocian wrote: > Hello, > >> I *think* this also fixes the VM hang reported in [2], where several >> processes in the guest were stuck in uninterruptible sleep. I am not >> familiar with the block layer, but my theory is that the threads were stuck >> in wait_for_completion_io(), from blk_execute_rq() executing a flush >> request. Milan has agreed to give this patch a spin [3], but that might >> take a while because the bug is not easily reproducible. I believe the >> patch can be merged on its own. >> >> [1] http://diy.inria.fr/www/index.html?record=aarch64&cat=aarch64-v04&litmus=SB%2Bdmb.sys&cfg=new-web >> [2] https://www.spinics.net/lists/kvm/msg204543.html >> [3] https://www.spinics.net/lists/kvm/msg222201.html >> > Unfortunately it didn't help. I can see the problem again now. Thank you for giving this a go. Unfortunately this means there's another bug lurking in kvmtool. I'll put this on my todo list and I'll try to figure something out when I have some spare time. Thanks, Alex
