[Openmcl-devel] armcl memory bug 1257 and v1.12-dev.4-3-gdd5622e9

[JTK]

Hello,

I’m running some multithreaded programs on an RPi 3b+, and 
I learned of memory corruption bug #1257: https://trac.clozure.com/ccl/ticket/1257 <https://trac.clozure.com/ccl/ticket/1257>
so I decided to compile the newest opencml to see if it is still there.

Strangely, the code (read-from-serialized-string) from the above page 
does not seem to trigger the bug after about an hour of running,
but my question is why is this version running on only 1 CPU?
And why does it seem much slower than before?

Top shows 100% cpu usage in  (read-from-serialized-string :thread-count 7),
but running the same code on 1.11.5/v1.11.5 uses nearly 400% CPU, so the new version 
1.12-dev seems to be running on one CPU.

Running dummy loops in separate threads in 1.12-dev uses up multiple CPUs, though,
so my 1.12-dev is definitely multithreaded.

So it’s only the task in ticket 1257 (or perhaps the lisp reader it uses?) that is single
threaded.

Can anyone shed some light on this?  Has this bug been effectively solved, or has
some unrelated change to threading simply mitigated it?

Thanks,
John


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.clozure.com/pipermail/openmcl-devel/attachments/20190110/9819e77c/attachment.htm>