[Fault-injection-developer] Fault Injection Test Harness Status Report

[Lynch, R. <rus...@in...>]

Fault Injection Test Harness Status Report
Date:  October 24, 2002

As a result of feedback from the Linux Kernel Mailing List (LKML) for a 
general need for a fault injection tool to aid kernel developers in finding 
bugs in the kernel, this project has experienced a boost over the last
couple 
of weeks.  A large part of this increase in activity has been a review of
the 
source code uploaded by Louis, Stanley, and Frank.

Several discussions have been sparked with the following changes agreed on:

* Kernel and user space code needs to be separated into two trees
  - It makes it much harder to follow when the two types of code are mixed
  - It makes it much harder to get our code reviewed by kernel developers
  - The kernel code needs to be ported to  the 2.5 kernel
  - In general Linux kernel development is happening on the 2.5 kernel.  
    If we hope to make a real contribution to Linux we need to be were the 
    development is happening.
  - We need to rethink our naming scheme
  - There is a proposal to make 'fith' only refer to the user space
utilities 
    that take advantage of fault injection features in the kernel
  - For kernel level code, a suggestion was made to use 'fi' to indicate 
    'fault injection'.  Some examples of where this would be used are:
    * 'fi_interceptor' to indicate kernel module that intercepts a specific 
      type of kernel action in order to inject a fault.

* We believe that the current state of the project would not be well
received 
  by LKML, so we looking over everything we have so far and targeting an
LKML 
  announcement by Christmas.


----------------------------------------------------------------------------
Rusty Lynch              
Staff Software Engineer,  Intel Corporation
rus...@in...

*  This email message solely contains my own personal views, and not 
   necessarily those of my employer