On Tue, Feb 05, 2019 at 09:57:50AM +0100, Peter Zijlstra wrote: > On Mon, Feb 04, 2019 at 03:24:23PM -0800, Dave Hansen wrote: > > > Actually, there's one part of all this that I forgot. Will split lock > > detection be enumerated _widely_? IOW, will my laptop in 5 years > > enumerate support for it? > > I would bloody hope so. Just for giggles, create an little program that > does LOCK prefix across a line or page boundary in a while(1) loop and > 'enjoy' your laptop experience. > > > If so, we surely don't want to enable this > > everyhwhere: it will break old apps. Doesn't that mean that we need > > both feature detection and another separate bit for folks to opt-in? > > No, we very much do want to default enable this everywhere. > > We might want to provide some chicken bits, like a (inheritable) PRCTL > or ELF flag to disable it for those broken apps.
Combined with a sysctl that disables the chickens; such that administrators can configure their system to be #AC pure. > But realistically, any app that triggers this is non-portable already, > most (if not all) the RISC architecture would already kill it with > SIGBUS for this. > > We very much want the kernel _AND_ firmware to be #AC clean, always, > everywhere. > > Heck, I'd love for #AC to be even stronger and not only trigger on > cross-line.
