Re: [PATCH v5 25/29] arm64: allow LTO_CLANG and THINLTO to be selected

[Will Deacon]

On Mon, Oct 12, 2020 at 01:44:56PM -0700, Kees Cook wrote:
> On Mon, Oct 12, 2020 at 09:31:16AM +0100, Will Deacon wrote:
> > On Fri, Oct 09, 2020 at 09:13:34AM -0700, Sami Tolvanen wrote:
> > > Allow CONFIG_LTO_CLANG and CONFIG_THINLTO to be enabled.
> > > 
> > > Signed-off-by: Sami Tolvanen <samitolvanen@xxxxxxxxxx>
> > > Reviewed-by: Kees Cook <keescook@xxxxxxxxxxxx>
> > > ---
> > >  arch/arm64/Kconfig | 2 ++
> > >  1 file changed, 2 insertions(+)
> > > 
> > > diff --git a/arch/arm64/Kconfig b/arch/arm64/Kconfig
> > > index ad522b021f35..7016d193864f 100644
> > > --- a/arch/arm64/Kconfig
> > > +++ b/arch/arm64/Kconfig
> > > @@ -72,6 +72,8 @@ config ARM64
> > >  	select ARCH_USE_SYM_ANNOTATIONS
> > >  	select ARCH_SUPPORTS_MEMORY_FAILURE
> > >  	select ARCH_SUPPORTS_SHADOW_CALL_STACK if CC_HAVE_SHADOW_CALL_STACK
> > > +	select ARCH_SUPPORTS_LTO_CLANG
> > > +	select ARCH_SUPPORTS_THINLTO
> > 
> > Please don't enable this for arm64 until we have the dependency stuff sorted
> > out. I posted patches [1] for this before, but I think they should be part
> > of this series as they don't make sense on their own.
> 
> Oh, hm. We've been trying to trim down this series, since it's already
> quite large. Why can't [1] land first? It would make this easier to deal
> with, IMO.

I'm happy to handle [1] along with the LTO Kconfig change when the time
comes, if that helps. I just don't want to merge dead code!

Will

> > [1] https://git.kernel.org/pub/scm/linux/kernel/git/will/linux.git/log/?h=rwonce/read-barrier-depends