[Mono-devel-list] patch as interp.c needs metadata_section initialized

[Bernie Solomon]

That seems right to me - just have one static CRITICAL_SECTION in interp.c (now mono_runtime_class_init grabs both this and the domain lock the deadlock which could happen before should have gone - I did have a different way around that but this is better).

Bernie Solomon

  ----- Original Message ----- 
  From: Laurent Morichetti 
  To: mono-devel-list at lists.ximian.com 
  Sent: Monday, July 21, 2003 5:09 PM
  Subject: Re: [Mono-devel-list] patch as interp.c needs metadata_section initialized


  Then, should calc_offset use metadata_section instead of metadata_lock?

  Bernie Solomon <bernard at ugsolutions.com> wrote: The interpreter isn't initializing metadata_section in the latest CVS code. I attach a patch to do so.

    Bernie Solomon
    Index: interp.c
    ===================================================================
    RCS file: /mono/mono/mono/interpreter/interp.c,v
    retrieving revision 1.244
    diff -u -p -r1.244 interp.c
    --- interp.c 20 Jul 2003 18:20:30 -0000 1.244
    +++ interp.c 21 Jul 2003 23:28:03 -0000
    @@ -4506,6 +4506,8 @@ quit_function (MonoDomain *domain, gpoin

    }

    +static CRITICAL_SECTION ms;
    +
    int 
    main (int argc, char *argv [])
    {
    @@ -4582,7 +4584,8 @@ main (int argc, char *argv [])
    mono_install_stack_walk (interp_walk_stack);
    mono_runtime_install_cleanup (quit_function);

    - InitializeCriticalSection (&metadata_lock);
    + metadata_section = &ms;
    + InitializeCriticalSection (metadata_section);
    domain = mono_init (file);
    mono_runtime_init (domain, NULL, NULL);

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.ximian.com/pipermail/mono-devel-list/attachments/20030721/4ed162ce/attachment.html