[Chicken-hackers] [PATCH] Fix #877, hopefully

[Peter Bex]

Hi all,

I think I finally figured out the problem of #877; because it's sending
alternate signals from threads, some signals were probably arriving
while the child process' handler was adding another signal to the queue.
This would in turn cause the position in the queue to get corrupted.

At least, I think that's what was happening...  Concurrency makes my
brain hurt.  I ran a few tests and I can't trigger the signal-test error
anymore.  But that was also the case with my previous patch, so it could
be dumb "luck".  Let's push it and see if Salmonella starts complaining.
It would also be great if everyone tests it on their multi-CPU boxes.

Mario told me that the "holes" in the output at
http://tests.call-cc.org/master/linux/x86/2014/02/ are due to this test
failing, so we should see it soon enough on the test server too.

Cheers,
Peter
-- 
http://www.more-magic.net

0001-Block-all-signals-while-the-global-signal-handler-is.patch
Description: Text document