Messages in this thread | | | From | Iago Abal <> | Date | Mon, 21 Oct 2013 12:49:32 +0200 | Subject | Looking for configuration-dependent bugs in Linux |
| |
Dear Linux kernel developer,
We are a group working on analysis of configurable software systems at IT University of Copenhagen. For the next four years [1] we will study the problem of verification in highly configurable software, and we will focus on the Linux kernel as the prime example of configurable software. We will therefore appreciate your advice regarding a few questions (see below).
We are currently looking for already-fixed configuration-dependent bugs in Linux ---the bugfix helps us to understand the underlying cause. A good example is [2], which fixes a bug that only shows up in PA-RISC architectures when DISCONTIGMEM is enabled. Any bug appearing in some configurations but not in others is interesting for us (e.g. a bug affecting PowerPC only), even though the more complex the interaction between configuration options the more interesting is the bug for our purpose.
As a Linux kernel developer, could you:
* remember any configuration-dependent bug? * identify any subset of configuration options that tend to interact badly? * spot any part of the kernel more prone to configuration-dependent bugs?
Your feedback will allow us to make our search more effective. A good understanding of how these bugs occur in Linux will also be important for developing subsequent techniques and tools. There is of course a potential benefit for the Linux community: we plan to pay you back in bug reports :-)
All your feedback is welcomed.
Thank you for your time.
Kind regards,
[1] http://variete.wikit.itu.dk/ [2] http://git.kernel.org/cgit/linux/kernel/git/stable/linux-stable.git/commit/?id=91ea8207168793b365322be3c90a4ee9e8b03ed4
-- Iago Abal Rivas - http://www.iagoabal.eu Ph.D. Fellow at IT University of Copenhagen
| |