On Fri, 31 Jul 2020 10:37:34 +0100
Daniel P. BerrangĂ© <[email protected]> wrote:

> On Fri, Jul 31, 2020 at 11:27:04AM +0200, Lukas Straub wrote:
> > yank_register_instance (called when creating the new chardev object)
> > aborts if the instance already exists. So check for duplicate id before
> > creating the new chardev to prevent this.  
> 
> I feel the right answer here is for yank_register_instance to not
> use abort().
> 
> Instead have it take a 'Error **errp' and report the error normally.
> The caller can then propagate the errors in the same way it does
> for the duplicate ID errors. If a particular caller can't handle
> errors gracefully, then it can pass "&error_abort" to the
> yank_register_instance() to get the same abort semantics as now.
> 
> Regards,
> Daniel

Yes, makes sense for yank_register_instance not to kill qemu. Although (with 
this fix in) it always indicates a bug due to a leaked yank instance.

Regards,
Lukas Straub

Attachment: pgprxCyl5OMqK.pgp
Description: OpenPGP digital signature

Reply via email to