On Thu, Feb 25, 2016 at 09:02:04AM +0100, Ingo Molnar wrote:
>
> * Josh Poimboeuf <jpoim...@redhat.com> wrote:
>
> > This is v17 of the compile-time stack metadata validation patch set.
> >
> > It's based on tip:x86/debug. However, note that when run against that
> > branch it will give a lot of warnings:
> >
> > objtool: arch/x86/ia32/sys_ia32.o: __ex_table size not a multiple of 12
> > objtool: arch/x86/ia32/ia32_signal.o: __ex_table size not a multiple of 12
> > objtool: arch/x86/entry/common.o: __ex_table size not a multiple of 12
> > ...
> >
> > These warnings means that objtool is expecting the new exception table
> > format which was introduced with:
> >
> > 548acf19234d ("x86/mm: Expand the exception table logic to allow new
> > handling options")
> >
> > So that commit is needed for those warnings to go away.
>
> Ok. I have created a new topic branch for the objtool commits:
> tip:core/objtool,
> and merged tip:ras/core into it (which hosts 548acf19234d).
>
> So please use tip:core/objtool as a base from now on.
>
> Two more minor observations:
>
> Please re-order the patches slightly so that the annotations are added first,
> and
> objtool is added and Kconfig-enabled in the final patches.
>
> > x86/objtool: Compile-time stack metadata validation
> > x86/objtool: Add CONFIG_STACK_VALIDATION option
> > x86/objtool: Enable objtool on x86_64
> > x86/objtool: Add STACK_FRAME_NON_STANDARD macro
> > x86/objtool: Add directory and file whitelists
>
> btw., please use the more generic 'objtool: ' prefix, the concept is not
> fundamentally x86 specific - especially once dwarf debuginfo is interpreted
> it
> will be useful beyond x86 as well I suspect.
>
> > x86/xen: Add xen_cpuid() to objtool whitelist
> > bpf: Add __bpf_prog_run() to objtool whitelist
> > sched: Add __schedule() to objtool whitelist
> > x86/kprobes: Add kretprobe_trampoline() to objtool whitelist
Ok. I'll make all your suggested changes and post v18 today.
--
Josh
