On Tue, Jan 09, 2018 at 10:46:50AM -0600, Victor Wren wrote:
> On 1/9/2018 09:02, Victor Wren wrote:
> > I'm stumped as to what switch enables stack validation and the ORC
> > unwinder that depends on libelf. The only way I've seen online to
> > disable stack validation is to manually edit the .config file.
>
> Looking through the Makefiles, if I am reading it at all correctly, X86_64
> automatically enables "CONFIG_HAVE_STACK_VALIDATION" which enables
> "CONFIG_STACK_VALIDATION" and the selection of CONFIG_UNWINDER_ORC
> automatically forces "CONFIG_STACK_VALIDATION" to be on.
>
> from arch/X86/Kconfig.debug:
> "config UNWINDER_ORC
> bool "ORC unwinder"
> depends on X86_64
> select STACK_VALIDATION"
>
> So that's where CONFIG_STACK_VALIDATION gets turned on. You have to either
> pick the ORC unwinder, or the slower, bulkier CONFIG_UNWINDER_FRAME_POINTER,
> that forces on CONFIG_FRAME_POINTER, but allows for building without libelf
> (I think). I still can't figure out if it's possible to disable all types
> of stack validation on x86_64. Seems that collecting debugging info on
> 64BIT is not optional?
> —Victor Wren
The difference with 32-bit x86 is that there only the old unwinder
is available. The unwinder is useful if a kernel bugs or oopses
(you'll probably need to post on lkml if that happens), and the new
Orc unwinder is intended to do a much better job - the old frame
pointer version often produces unhelpful results.
ĸen
--
Truth, in front of her huge walk-in wardrobe, selected black leather
boots with stiletto heels for such a barefaced truth.
- Unseen Academicals
--
http://lists.linuxfromscratch.org/listinfo/lfs-dev
FAQ: http://www.linuxfromscratch.org/faq/
Unsubscribe: See the above information page
