Re: [PATCH] Add more rules to docs/misra/rules.rst

[Stefano Stabellini <sstabellini@xxxxxxxxxx>]

On Mon, 30 Jan 2023, Luca Fancellu wrote:
> > On 30 Jan 2023, at 07:33, Jan Beulich <jbeulich@xxxxxxxx> wrote:
> > 
> > On 27.01.2023 19:33, Stefano Stabellini wrote:
> >> On Fri, 27 Jan 2023, Jan Beulich wrote:
> >>> On 26.01.2023 19:54, Stefano Stabellini wrote:
> >>> Looking back at the sheet, it says "rule already followed by
> >>> the community in most cases" which I assume was based on there being
> >>> only very few violations that are presently reported. Now we've found
> >>> the frame_table[] issue, I'm inclined to say that the statement was put
> >>> there by mistake (due to that oversight).
> >> 
> >> cppcheck is unable to find violations; we know cppcheck has limitations
> >> and that's OK.
> >> 
> >> Eclair is excellent and finds violations (including the frame_table[]
> >> issue you mentioned), but currently it doesn't read configs from xen.git
> >> and we cannot run a test to see if adding a couple of deviations for 2
> >> macros removes most of the violations. If we want to use Eclair as a
> >> reference (could be a good idea) then I think we need a better
> >> integration. I'll talk to Roberto and see if we can arrange something
> >> better.
> >> 
> >> I am writing this with the assumption that if I could show that, as an
> >> example, adding 2 deviations reduces the Eclair violations down to less
> >> than 10, then we could adopt the rule. Do you think that would be
> >> acceptable in your opinion, as a process?
> > 
> > Hmm, to be quite honest: Not sure. Having noticed the oversight of the
> > frame_table[] issue makes me wonder how much else may be missed in this
> > same area (18.1, 18.2, and 18.3).
> 
> Hi Jan,
> 
> I think I recall the frame_table[] issue but I was looking into the eclair 
> reports to
> understand it better and I was unable to find it, do you recall where the 
> tool was
> complaining for the 18.2 related to the frame_table[]?
> Any notes or link is appreciated, maybe we could speak with Roberto to 
> understand
> It better, because I checked with Coverity and I was unable to link findings 
> of 18.2 with
> the symbol frame_table[] (however I might be a bit lost in all the macros).

Eclair could find the following violation which is related to the
frametable in arch/x86/mm.c:init_frametable_chunk:

   L229: memset(start, 0, end - start);

https://eclairit.com:8443/job/XEN/lastBuild/Target=X86_64,agent=docker1/eclair/type.-1600986843/moduleName.-84949685/packageName.590865259/fileName.1553859513/
https://eclairit.com:3787/fs/var/lib/jenkins/jobs/XEN/configurations/axis-Target/X86_64/axis-agent/docker1/builds/686/archive/ECLAIR/out/PROJECT.ecd;/sources/xen/arch/x86/mm.c.html#R46776_1

Almost all the other 18.2 violations are due to the "guest_mode" macro
and the NEEP_IP/OP HAVE_IO/OP macros in lzo.c.