On Mon, Mar 06, 2017 at 04:20:13PM +0100, Peter Zijlstra wrote: > On Mon, Mar 06, 2017 at 03:24:23PM +0100, Dmitry Vyukov wrote: > > We could also provide a parallel implementation of atomic ops based on > > the new compiler builtins (__atomic_load_n and friends): > > https://gcc.gnu.org/onlinedocs/gcc/_005f_005fatomic-Builtins.html > > and enable it under KSAN. The nice thing about it is that it will > > automatically support arm64 and KMSAN and KTSAN. > > But it's more work. > > There's a summary out there somewhere, I think Will knows, that explain > how the C/C++ memory model and the Linux Kernel Memory model differ and > how its going to be 'interesting' to make using the C/C++ builtin crud > with the kernel 'correct. Trivially, The C++ model doesn't feature I/O ordering [1]... Otherwise Will pointed out a few details in [2]. Thanks, Mark. [1] https://lwn.net/Articles/698014/ [2] http://lwn.net/Articles/691295/ -- To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to majordomo@xxxxxxxxx. For more info on Linux MM, see: http://www.linux-mm.org/ . Don't email: <a href=mailto:"dont@xxxxxxxxx";> email@xxxxxxxxx </a>
