Re: [PATCH 1/4] mm/kasan: get rid of speculative shadow checks

2017-06-01 Thread Dmitry Vyukov
On Thu, Jun 1, 2017 at 6:23 PM, Andrey Ryabinin wrote: > For some unaligned memory accesses we have to check additional > byte of the shadow memory. Currently we load that byte speculatively > to have only single load + branch on the optimistic fast path. > > However, this approach have some downs

[PATCH 1/4] mm/kasan: get rid of speculative shadow checks

2017-06-01 Thread Andrey Ryabinin
For some unaligned memory accesses we have to check additional byte of the shadow memory. Currently we load that byte speculatively to have only single load + branch on the optimistic fast path. However, this approach have some downsides: - It's unaligned access, so this prevents porting KASAN on