#1 thread class: suspend/resume error

Common_Libraries
closed
David Sugar
Library (32)
3
2005-06-06
2005-05-13
Nicola Bellotto
No

this bug was already reported in the previous project
website:
the method suspend() stops the process and a following
resume() doesn't work.
this happens for sure on linux fedora 3.
below there's the output of the thread1.cpp example.

***********************************************
* Testing class Thread without syncronization *
***********************************************
Testing thread creation

- thread should set n to 1...ok

Testing thread is working

- thread should change n...ok
- thread should change n...ok

Testing suspend & resume

[1]+ Stopped t1

Discussion

  • David Sugar
    David Sugar
    2005-05-15

    • labels: --> Library
    • milestone: --> Common_Libraries
    • priority: 5 --> 3
    • assigned_to: nobody --> dyfet
     
  • David Sugar
    David Sugar
    2005-05-15

    Logged In: YES
    user_id=217

    Yes, there are quirky methods used to try and simulate the behavior of
    thread suspend, many are based on signals. In this case, the method used
    on the gnu/linux targets worked on earlier linux kernels/glibc's when threads
    were still seperate processes, and hence the process suspend signal was
    actually thread local. A new and different hack will probably need to be
    created for newer pthread implimentations. Some work was done around
    other signal methods with sigwait, but these behaviors are also rather
    inconsistent.

     
  • David Sugar
    David Sugar
    2005-06-06

    • status: open --> closed