Home | History | Annotate | Line # | Download | only in roadmaps
      1 $NetBSD: verification,v 1.2 2019/06/17 17:45:18 kamil Exp $
      2 
      3 NetBSD Verification Roadmap
      4 ===========================
      5 
      6 This roadmap covers things that we intend or would like to do pursuant
      7 to verification and quality control.
      8 
      9 The following elements, projects, and goals are relatively near-term:
     10 
     11  1. Cut down the Coverity backlog
     12  3. Deploy clang-static-analyzer
     13 
     14 The following elements, projects, and goals are longer-term:
     15 
     16  4. mint
     17  5. Database-driven program analyzer
     18 
     19 The following elements, projects, and goals are rather blue-sky so far:
     20 
     21  6. Use Frama-C to verify fsck_ffs
     22 
     23 
     24 Explanations
     25 ============
     26 
     27 1. Cut down the Coverity backlog
     28 
     29 Coverity provides us with free analysis reports, which we sometimes
     30 handle and sometimes don't. Apparently though Linux has a lower number
     31 of Coverity hits per line than we do; that seems fundamentally wrong
     32 and something that should get attention. Most of the problems Coverity
     33 finds are pretty easily fixed, or are false positives.
     34 
     35  - As of January 2017 coypu has been working on this. Christos often
     36    also fixes these.
     37  - There is currently no clear timeframe or release target.
     38  - Contact christos for further information.
     39 
     40 
     41 3. Deploy clang-static-analyzer
     42 
     43 There is some makefile support for running clang-static-analyzer, but
     44 it doesn't get done regularly. This should probably get added to the
     45 autobuilds.
     46 
     47  - As of January 2017 nobody is known to be working on this.
     48  - There is currently no clear timeframe or release target.
     49  - Contact joerg for further information.
     50 
     51 
     52 4. mint
     53 
     54 A while back dholland started on a replacement for the existing lint,
     55 since lint is not really smart enough to be useful and its code is
     56 only marginally maintainable. The code is in othersrc, but it needs
     57 some tidying before anyone else tries hacking on it.
     58 
     59  - As of January 2017 nobody is known to be working on this.
     60  - There is currently no clear timeframe or release target.
     61  - Responsible: dholland
     62 
     63 
     64 5. Database-driven program analyzer
     65 
     66 In the long run we would like to have a program analyzer that can
     67 scale to the whole kernel and can do differential analyses across
     68 different versions. This is a nontrivial project though.
     69 
     70  - As of January 2017 nobody is known to be working on this.
     71  - There is currently no clear timeframe or release target.
     72  - Contact: dholland
     73 
     74 
     75 6. Use Frama-C to verify fsck_ffs
     76 
     77 Frama-C is a framework for doing formal verification of C code using
     78 (mostly) precondition/postcondition specs. It is not everything one
     79 necessarily wants in a verification framework; but on the other hand
     80 it exists and people do use it.
     81 
     82 fsck_ffs seems like a good candidate for this because it's
     83 mission-critical and what it needs to do is comparatively well
     84 understood even in detail. However, the code may be too messy.
     85 
     86  - As of January 2017 nobody is known to be working on this.
     87  - There is currently no clear timeframe or release target.
     88  - Contact: dholland
     89