Re: aborting IO requests

Note: the cbdata usage of sio objects is a little bit messy with no
dedicated "owner", partly due to dual events that may cause a sio to
be closed down making it hard to define who is the "owner".

  * storeClose()
  * I/O errors

. Should probably be changed to make the core the logical cbdata
"owner" of the object (who is the only one who is allowed to trigger
a cbdataFree() directly or indirectly by design)

Regards
Henrik

On Tuesday 03 December 2002 07.03, Henrik Nordstrom wrote:
> sio is a cbdata. cbdata things are aborted by cbdataFree().
>
> For storeio the abort operation is piggy backed on storeClose(sio)
> -> SD.obj.close(sio).
>
> * Pending reads are aborted if possible
> * Pending writes are completed
>
>
> In the long run there needs to be a specific abort file operation
> added to the "fs" API me thinks.
>
> Regards
> Henrik
>
> On Tuesday 03 December 2002 05.36, Robert Collins wrote:
> > I can't find the code to trigger this!
> >
> > I suspected it to be a cbdataReferenceDone(foo->sio), but I can't
> > find such a statement.
> >
> > Henrik,
> > you were looking near this area not so long ago, care to
> > comment?
> >
> > Rob
Received on Mon Dec 02 2002 - 23:38:54 MST