Mir client library thread safety

[Alan Griffiths]

On 24/05/13 05:37, Thomi Richards wrote:
> Hi Folks,
>
>
> I found another thread safety issue in mir - this time in the client
> library. There's a detailed bug reported here:
>
> https://bugs.launchpad.net/mir/+bug/1183659
>
> I started fixing the code myself, but My C/C++ is a little rusty. My
> fix is here:
>
> https://code.launchpad.net/~thomir/mir/fix-client-thread-race
> <https://code.launchpad.net/%7Ethomir/mir/fix-client-thread-race>
>
> The fix seems to work (for my mir-stress test suite anyway), but I had
> problems fixing the failing tests.
>
> It's now end-of-week for me, so I'd appreciate it if someone from the
> US or Europe could take a look at the bug and my branch and confirm
> that I've correctly identified the issue, and perhaps shepherd a fix
> into trunk.


AFAICS There's only one problem test (which hangs) - and the reason it
isn't easy to fix is that the original logic is broken too.

Investigating.