This series fixes a race condition that occurs under OS X. It also reworks the signal initialization to make it simpler for later maintenance/additions.
Note that although it _appears_ to fix this race condition, I have not been able to pinpoint exactly how it is triggered. Does anyone know if there is a specific reason why SIGUSR2 is not currently handled via the signal thread? Alexandre Raymond (2): Signals: fix race condition with aio-compat Signals: rework initial signal setup cpus.c | 43 +++++++++++++++++++++++-------------------- 1 files changed, 23 insertions(+), 20 deletions(-) -- 1.7.5
