[patch] Add a level of abstraction to condvars and mutexes so we can initalize them late

John (J5) Palmieri johnp at redhat.com
Tue Aug 15 20:38:13 PDT 2006


New patch with test case.  Looking at the dataslots again and I see your
point about the global locks.  

On Tue, 2006-08-15 at 13:09 -0400, Havoc Pennington wrote:
> John (J5) Palmieri wrote:
> > Here is the patch.  I need to write a test case but perhaps other may
> > spot something I missed.  One question on my mind is if the data-slots
> > API needs to use the extra overhead of the abstraction because it seems
> > to use mutexes in a very self contained way. i.e. the _dbus_mutex_new is
> > freed at the end of the function it is called in and the rest of the
> > D-Bus API uses data-slots with global locks that don't need to be
> > abstracted.
> 
> Why do you need _at_location for anything except new/free ? the 
> lock/unlock/wait methods should not need this afaict.
> 
> Not sure I understand the question about data slots. They are using 
> global mutexes, but you do have to replace the reference to the global 
> mutex in the data slot allocator object, if you replace the global mutex 
> itself.
> 
> Havoc
> 
-------------- next part --------------
A non-text attachment was scrubbed...
Name: dbus-init-threads-late-2.patch
Type: text/x-patch
Size: 41137 bytes
Desc: not available
Url : http://lists.freedesktop.org/archives/dbus/attachments/20060815/f7414f57/dbus-init-threads-late-2-0001.bin


More information about the dbus mailing list