> > [ Upstream commit 706d51681d636a0c4a5ef53395ec3b803e45ed4d ] > > > > Changes from upstream: > > ---------------------- > > 1. Use bit 30 of word 7 in cpufeatures for X86_FEATURE_IBRS_ENHANCED as > bit 29 > > is now used by L1TF. > > 2. Fix some trivial line fuzzing. > > > > Based on kernel version: > > ------------------------ > > Linux 4.9.135 > > This format is very odd, shouldn't it go all below the original --- line? > > Please fix up and resend properly. Sure! My bad, sorry! As suggested, I will fix this up and will resend a V2. > Also, why 4.9? What is wrong with 4.18.y, > 4.14.y, and 4.4.y? Just picking one random kernel in the middle of the stable > releases seems foolish for those users of newer kernels, right? I have skipped 4.4 because, 1. It's too old 2. x86/kvm/vmx.c doesn't use x86_spec_ctrl_set_guest() and x86_spec_ctrl_restore_host() before entering guest and after exiting guest and entering host. This is needed because IBRS bit in SPEC_CTRL register could be modified by guest but host should still have it set always. Regards, Sai
