On 13/03/2018 14:20, Vitaly Kuznetsov wrote: > vmcs_writel() I guess ... and, just to make sure I follow your > suggestion, this is for x86_64 only, right? x86_32 does > > vmcs_writel(HOST_FS_BASE, segment_base(vmx->host_state.fs_sel)); > > and I think it needs to stay.
Yes. > (personally, I'm rather for exporting save_fsgs(), dropping > savesegment() and getting all we need from current to avoid propagating > assumptions but I'm flexible) Yes, that's fine too, as long as it's not mix and match. :) Paolo