Re: [XEN PATCH v2] automation/eclair_analysis: deviate linker symbols for Rule 18.2

[Stefano Stabellini <sstabellini@xxxxxxxxxx>]

On Tue, 10 Sep 2024, Jan Beulich wrote:
> On 10.09.2024 10:18, Nicola Vetrini wrote:
> > On 2024-09-10 08:26, Jan Beulich wrote:
> >> On 10.09.2024 06:46, Stefano Stabellini wrote:
> >>> On Mon, 9 Sep 2024, Jan Beulich wrote:
> >>>> On 07.09.2024 15:03, Nicola Vetrini wrote:
> >>>>> +   * - R18.2
> >>>>> +     - Subtraction between pointers encapsulated by macro 
> >>>>> page_to_mfn
> >>>>> +       are safe.
> >>>>> +     - Tagged as `safe` for ECLAIR.
> >>>>
> >>>> This one is a result of using frame_table[], aiui. Alternative 
> >>>> approaches
> >>>> were discussed before. Did that not lead anywhere, requiring a purely
> >>>> textual / configurational deviation?
> >>>
> >>> During the last MISRA discussion we agree that this was an acceptable
> >>> approach. What else did you have in mind?
> >>
> >> One was to have the linker scripts provide the symbol. I think there 
> >> were
> >> one or two more, yet I - perhaps wrongly - haven't been taking notes 
> >> ...
> >>
> > 
> > One thing I'm fairly sure has been suggested for symbols that were not 
> > linker-defined is the following mitigation:
> > 
> > gcc -fsanitize=address,pointer-subtract
> > ASAN_OPTIONS=detect_invalid_pointer_pairs=2 ./a.out
> 
> ASAN in general was mentioned in the discussion, yet it didn't look like
> anyone would be up to actually making ASAN usable with Xen. Iirc Andrew
> estimated the effort to a man-year.

Given this, let's stay with the patch and deviation in the current form