I'm a member of the metacompilation research group at Stanford (http://www.stanford.edu/~engler). We have a suite of checkers that find bugs at compile time, and we've had quite a bit of success checking the Linux kernel code for errors. Since our checkers can emit false alarms, we filter the reports before we give them to the kernel developers. While some false alarms slip past us to the developers, our limited knowledge of the kernel allows us to recognize most of them.
We're currently trying to find race conditions and deadlock (here's an example report to the kernel mailing list: http://www.ussg.iu.edu/hypermail/linux/kernel/0303.2/1733.html), but these reports seem to require an intimate knowledge of the code to verify-- knowledge that we don't have. Since the kernel is so large, it is difficult to find the various developers who know the code involving the reports. As a result, many of these bug reports go unconfirmed even though we think they are valid bugs. I'm hoping to find another project to supplement our race condition and deadlock work on the Linux kernel, and I think that Apache might be such a project. So I have some questions for you all to determine if this is a worthwhile venture: Have race conditions and deadlock been a problem in the past? How likely is it that there are race condition and deadlock bugs hiding in the current source? Who are the developers who could answer my "is this a race condition" questions? Is there any documentation about locks in the server? Where they are used? How they are used? What do they protect? What files should I be looking at? Which use locks? Which contain the locking functions? Are there any absolute rules about locks (i.e. all global variables must be protected by locks, orderings of lock acquisition)? Thanks for your time, Ken Ashcraft [EMAIL PROTECTED]
