On Tue, 2015-03-03 at 22:45 +0100, Xavier Chantry wrote:
> Hello,
>
> Using Debian 3.16.7-ckt4-3 and a simple test case, we were able to
> reproduce a kernel bug in msync system call.
> I could try to rewrite the test case using standard pthread if
> necessary, but it is quite basic and does the following:
> 1) create a 1MB file and mmap it
> 2) in a loop, select a 4k page in that file randomly, write with
> memset, then run msync on that area in a thread
>
>
> It appears that concurrent msync trigger very easily the following bug
> using debian 3.16 kernel. It was first seen with debian 3.12, and
> debian 3.2 was fine.
> Using a vanilla kernel with a custom config, I've never been able to
> reproduce the issue (tested on 3.17 3.18 3.19 kernels).
>
> If you have any idea how to efficiently get to the bottom of this
> issue, let me know. If you just want me to open a bug on the debian
> bug tracker, also let me know :)
[...]
Please do.
Ben.
--
Ben Hutchings
Horngren's Observation:
Among economists, the real world is often a special case.
Attachment:
signature.asc
Description: This is a digitally signed message part