[Bug 152] New: thin lock algorithm probably needs barriers for SMP

http://sablevm.org/bugs/show_bug.cgi?id=152

           Summary: thin lock algorithm probably needs barriers for SMP
           Product: SableVM
           Version: trunk
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: default
        AssignedTo: [email protected]
        ReportedBy: [email protected]
         QAContact: [email protected]
                CC: [email protected]


This isn't specified by the new JMM, but Section 3.4 of Onodera's paper _A study
of locking objects with bimodal fields_ talks about potential problem with his
thin lock algorithm on MP systems.  I think we need these in SableVM as well,
since the thin lock algorithm SableVM uses is derived from Onodera's.

http://www.research.ibm.com/trl/projects/jit/paper/oopsla99_onodera.pdf

I have made conservative changes that I think are appropriate in my sandbox
(developer chris) in thread.c, along with lots of comments relating steps in the
thin lock algorithm back to the literature, but since SMP support is currently
broken for several reasons, it's hard to verify if they actually do anything.



------- You are receiving this mail because: -------
You are the QA contact for the bug, or are watching the QA contact.