On Mon, 2024-11-04 at 08:11 -0800, Dave Hansen wrote: > On 11/4/24 00:58, Shah, Amit wrote: > > Right - thanks, I'll have to reword that to say the RSB is flushed > > along with the TLB - so any action that causes the TLB to be > > flushed > > will also cause the RSB to be flushed. > > Hold on though. > > Is there a need for the RSB to be flushed at context switch? You > talked > about it like there was a need: > > > any hardware TLB flush results in flushing of the RSB (aka RAP in > > AMD spec). This guarantees an RSB flush across context switches. I want to justify that not setting X86_FEATURE_RSB_CTXSW is still doing the right thing, albeit in hardware.
