> It seems like an obvious fix. Both call traces look identical. I can
> certainly produce a test kernel and get the reporter to produce some
> results.
Just to be clear, both call traces look identical because I wrote that patch to fix this particular issue, but unfortunately didn't get it backported to the stable releases.
I'm not sure why we haven't seen it on trusty. Perhaps some initialization ordering or kernel config options changed which exposed the race.
> It seems like an obvious fix. Both call traces look identical. I can
> certainly produce a test kernel and get the reporter to produce some
> results.
Just to be clear, both call traces look identical because I wrote that patch to fix this particular issue, but unfortunately didn't get it backported to the stable releases.
I'm not sure why we haven't seen it on trusty. Perhaps some initialization ordering or kernel config options changed which exposed the race.