Re: [Xen-devel] [PATCH 7/9] libxl: Make killing of device model asynchronous

[Ian Jackson <ian.jackson@xxxxxxxxxx>]

George Dunlap writes ("Re: [PATCH 7/9] libxl: Make killing of device model 
asynchronous"):
> On Nov 28, 2018, at 4:43 PM, Ian Jackson <ian.jackson@xxxxxxxxxx> wrote:
> > Conversely it would be nice to say somewhere
> > that ddms->callback may be called reentrantly.
> 
> What do you mean by reentrantly?  That libxl__destroy_device_model() may end 
> up calling it directly (on this execution stack), rather than arranging for 
> it to be called by someone else after returning?

Precisely.

This kind of callback can, in some situations, be a hazard for the
caller, because it might result in data structures that the caller is
about to rely on vanishing.

It is conventional in libxl to return after setting up callbacks so
this generally doesn't arise, but adding a note to functions with this
property is a good idea.

Ian.

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxxx
https://lists.xenproject.org/mailman/listinfo/xen-devel