> On Apr 2, 2020, at 1:07 PM, Thomas Gleixner <tglx@xxxxxxxxxxxxx> wrote: > > > > TBH, the more I learn about this, the more I tend to just give up on > this whole split lock stuff in its current form and wait until HW folks > provide something which is actually usable: > > - Per thread > - Properly distinguishable from a regular #AC via error code Why the latter? I would argue that #AC from CPL3 with EFLAGS.AC set is almost by construction not a split lock. In particular, if you meet these conditions, how exactly can you do a split lock without simultaneously triggering an alignment check? (Maybe CMPXCHG16B? > > OTOH, that means I won't be able to use it before retirement. Oh well. > > Thanks, > > tglx
