Re: Test-lock hang (not 100% reproducible) on GNU/Linux

[Bruno Haible]

> > What happens when a program reads from a 'volatile' variable
> > at address xy in a multi-processor system?

Looking at the assembler code produced by GCC: GCC does not emit
any barrier or similar instructions for reads or writes to 'volatile'
variables. So, the loop in test_lock actually waits until the other
CPU _happens_ to flush it cache.

> Might be NUMA related?

Indeed. Probably the CPUs in a NUMA system flush their caches to
main memory not so frequently.

Whereas when we use a lock, the thread library or kernel executes
the appropriate cache flush instructions.

Bruno