Re: [PATCH 1/2] Add RUNTIME_DISCARD_EXIT to generic DISCARDS

From: Kees Cook
Date: Thu Mar 26 2020 - 16:39:08 EST


On Thu, Mar 26, 2020 at 09:11:42PM +0100, Borislav Petkov wrote:
> On Thu, Mar 26, 2020 at 12:30:20PM -0700, H.J. Lu wrote:
> > In x86 kernel, .exit.text and .exit.data sections are discarded at
> > runtime, not by linker. Add RUNTIME_DISCARD_EXIT to generic DISCARDS
> > and define it in x86 kernel linker script to keep them.
> >
> > Signed-off-by: H.J. Lu <hjl.tools@xxxxxxxxx>
> > Reviewed-by: Kees Cook <keescook@xxxxxxxxxxxx>
> > ---
> > arch/x86/kernel/vmlinux.lds.S | 1 +
> > include/asm-generic/vmlinux.lds.h | 10 ++++++++--
> > 2 files changed, 9 insertions(+), 2 deletions(-)
> >
> > diff --git a/arch/x86/kernel/vmlinux.lds.S b/arch/x86/kernel/vmlinux.lds.S
> > index e3296aa028fe..7206e1ac23dd 100644
> > --- a/arch/x86/kernel/vmlinux.lds.S
> > +++ b/arch/x86/kernel/vmlinux.lds.S
> > @@ -21,6 +21,7 @@
> > #define LOAD_OFFSET __START_KERNEL_map
> > #endif
> >
> > +#define RUNTIME_DISCARD_EXIT
> > #define EMITS_PT_NOTE
> > #define RO_EXCEPTION_TABLE_ALIGN 16
> >
> > diff --git a/include/asm-generic/vmlinux.lds.h b/include/asm-generic/vmlinux.lds.h
> > index e00f41aa8ec4..6b943fb8c5fd 100644
> > --- a/include/asm-generic/vmlinux.lds.h
> > +++ b/include/asm-generic/vmlinux.lds.h
> > @@ -894,10 +894,16 @@
> > * section definitions so that such archs put those in earlier section
> > * definitions.
> > */
> > +#ifdef RUNTIME_DISCARD_EXIT
> > +#define EXIT_DISCARDS
> > +#else
> > +#define EXIT_DISCARDS \
> > + EXIT_TEXT \
> > + EXIT_DATA
> > +#endif
>
> /me goes back and reads the old thread on this...
>
> Kees, do you expect other arches to actually need this
> RUNTIME_DISCARD_EXIT thing or was that a hypothetical thing?
>
> /me searches more...
>
> oh, there's a patchset from you
>
> https://lkml.kernel.org/r/20200228002244.15240-1-keescook@xxxxxxxxxxxx
>
> which already contains this patch *and* an ARM64 patch which defines
> RUNTIME_DISCARD_EXIT so I'm guessing ARM64 wants to discard at runtime
> too.

Correct.

> Which leaves the question why is H.J. sending that patch separate and
> you carry it in a patchset about orphan section warning? Seems like it
> wants to be in your patchset?

I had needed the same clean up for the orphan section handling, and
since it hadn't been picked up yet, I included it in my series. I'm
still stuck addressing several review comments, so there's no reason to
wait for me: I can easily rebase once these patches land somewhere. I'd
be happy to see them in -tip.

Thanks!

-Kees

--
Kees Cook