Re: [PATCH 7/7] gnttab: unreachable code when GNTTAB_MAX_VERSION < 2
Nicola Vetrini <[email protected]> Tue, 28 Jul 2026 17:53:52 +0200
| Newsgroups | gmane.comp.emulators.xen.devel |
|---|---|
| Organization | BUGSENG s.r.l. |
| Message-ID | <[email protected]> |
On 2026-07-28 17:42, Jan Beulich wrote: > On 28.07.2026 17:40, Nicola Vetrini wrote: >> On 2026-07-28 15:53, Jan Beulich wrote: >>> I'm surprised Eclair doesn't spot the large chunks of unreachable >>> code >>> on >>> Arm, i.e. violations of Misra C:2012 rule 2.1. >>> >>> Signed-off-by: Jan Beulich <[email protected]> >>> --- >>> This pretty certainly isn't dealing with everything. For example, >>> with >>> another helper the gt_version field could likely also become >>> conditional. >>> With some more effort the nr_status_frames field similarly could >>> become >>> conditional. >>> >> [...] >> >> Hard to judge only looking at the diff. Is this code unreachable in >> the >> arm64-allcode analysis configuration as well? > > Arm uniformly has > > #define GNTTAB_MAX_VERSION 1 > > so all gnttab-v2 code is unreachable there, no matter what > configuration. > > Jan I see. Existing deviations may play a role here in excluding caution reports that would otherwise be generated, particularly -config=MC3A2.R2.1,+reports={safe,"first_area(^.*has an invariantly.*$)"} I see only one condition "GNTTAB_MAX_VERSION" is used directly, while the others are via "opt_gnttab_max_version" which is why I reasoned there might be cautions involved, instead of violations. -- Nicola Vetrini, B.Sc. Software Engineer BUGSENG (https://bugseng.com) LinkedIn: https://www.linkedin.com/in/nicola-vetrini-a42471253