Ingo Molnar wrote: > * Avi Kivity <avi at qumranet.com> wrote: > > >>> my T60 laptop does not resume correctly due to KVM attempting to >>> send an IPI to a CPU that might be down (or not up yet). (Doing so >>> also triggers the send_IPI_mask_bitmask() warning in >>> arch/i386/kernel/smp.c, line 732.) >>> >>> with this fix applied my laptop does not hang during resume. >>> >>> [ KVM will have to disable/enable virtualization on the CPU itself >>> that goes down / comes up, not via an IPI sent from the requesting >>> CPU. ] >>> > > >> That is already CPU_ONLINE in my tree (and in the pull request sent to >> Linus a couple of days ago). >> > > that solves the resume problem - but doesnt solve the CPU_DEAD issue of > sending an IPI to an already offline CPU. Might be a better idea to do > it in CPU_DOWN_PREPARE? (and then to also add a CPU_DOWN_FAILED branch?) > > Mainline now has DOWN_PREPARE and UP_CANCELED calling ->hardware_disable(), and ONLINE calling ->hardware_enabled(). What tree are you looking at? [but I do see the need for DOWN_FAILED now. Off to find a resumable machine...] -- Do not meddle in the internals of kernels, for they are subtle and quick to panic.
