Re: [RFA] Use observers to report stop events.

[Vladimir Prus <vladimir@codesourcery.com>]

On Friday 02 May 2008 00:11:04 Pedro Alves wrote:
> A Thursday 01 May 2008 20:57:58, Daniel Jacobowitz wrote:
> 
> > Am I correct that the cleanup for finish_command is never supposed to
> > survive the function returning?  It's run on error and discarded on
> > normal return.  So you could put the closure in a local variable,
> > maybe.
> >
> >   struct foo_closure my_closure = { &my_global, my_global };
> >   make_cleanup (restore_integer, &my_closure);
> 
> Humm, I notice that a cleanup is being passed around to
> finish_command_continuation, but it isn't being used inside -- and
> it shoud not be.  With continuations per thread, and and non-stop, we
> can have more than one simultaneous finish command.  The cleanup
> chains are not per-thread, so it is not safe to run cleanups like that.
> Either we have to use some other form of cleanup in this continuation,
> or revert to having exec_cleanups, but this time, per thread.

We just should not pass cleanup to continuation -- as it's not used and
should not be. I've missed that bit in previous patches.

- Volodya