[Vm-dev] [OpenSmalltalk/opensmalltalk-vm] squeak.cog.v3 CI tests fail randomly (#436)

[Nicolas Cellier]

So, there was a  thread from March 2017 explaining the pitfalls in `primitiveSignalAtMilliseconds`:

> Re: [Vm-dev] We need help from VM experts. Re: Freeze after Morph Activity
http://lists.squeakfoundation.org/pipermail/vm-dev/2017-March/024381.html

In particular, all was said here: the rollover workaround is **HIGHLY suspiscious**:
http://lists.squeakfoundation.org/pipermail/vm-dev/2017-March/024410.html

So we solved it, or IMO we didn't solve, just patched:
http://lists.squeakfoundation.org/pipermail/vm-dev/2017-March/024438.html

As I later reminded, there were still problems:
http://lists.squeakfoundation.org/pipermail/vm-dev/2017-May/025160.html

But our attention was focused on more important developments (spur,win64 etc...) and we patched and replaced `usqLong deltaMsecs` by `sqLong deltaMsecs`, about 1 year after 1st problem:
http://lists.squeakfoundation.org/pipermail/vm-dev/2018-March/027041.html

The compiler warning disappeared, but we kept the rollover that I once suggested to remove!
We eliminate all smells of the problem, but not really the problem itself, that remain burried and quiet for some times (that's expected, squeak.cog.v3 might be used occasionally or less).
I'm ready to bet a few hundred $ that we can can still trigger a problem in cog.v3 as Juan did once, and that our CI failures are related. I haven't got an old image handy near this keyboard, but if anyone wants to try this in a squeak.cog.v3, and watchout what the console delivers:

    s := Semaphore new.
    Delay primSignal: s atMilliseconds: Time primMillisecondClock - 2.
    s wait.



-- 
You are receiving this because you are subscribed to this thread.
Reply to this email directly or view it on GitHub:
https://github.com/OpenSmalltalk/opensmalltalk-vm/issues/436#issuecomment-546430626
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.squeakfoundation.org/pipermail/vm-dev/attachments/20191025/27eb65e4/attachment.html>