On Mon, Jan 11, 2021 at 08:03:29PM +0100, Andrey Konovalov wrote: > On Mon, Jan 11, 2021 at 7:59 PM Nathan Chancellor > <natechancellor@xxxxxxxxx> wrote: > > > > > > -config KASAN_STACK_ENABLE > > > > +config KASAN_STACK > > > > bool "Enable stack instrumentation (unsafe)" if CC_IS_CLANG && !COMPILE_TEST > > > > > > Does this syntax mean that KASAN_STACK is only present for > > > CC_IS_CLANG? Or that it can only be disabled for CC_IS_CLANG? > > > > It means that the option can only be disabled for clang. > > OK, got it. > > > > Anyway, I think it's better to 1. allow to control KASAN_STACK > > > regardless of the compiler (as it was possible before), and 2. avoid > > > > It has never been possible to control KASAN_STACK for GCC because of the > > bool ... if ... syntax. This patch does not change that logic. Making it > > possible to control KASAN_STACK with GCC seems fine but that is going to > > be a new change that would probably be suited for a new patch on top of > > this one. > > The if syntax was never applied to KASAN_STACK, only to > KASAN_STACK_ENABLE, so it should have been possible (although I've > never specifically tried it). CONFIG_KASAN_STACK was not a user selectable symbol so it was always 1 for GCC. Cheers, Nathan
