On 22.12.2022 10:55, Demi Marie Obenour wrote: > On Thu, Dec 22, 2022 at 10:48:02AM +0100, Jan Beulich wrote: >> Also wasn't there talk of having this behavior in debug hypervisors >> only? Without that restriction I'm yet less happy with the change ... > > My understanding was that Andrew preferred the behavior to be turned on > in release hypervisors too. I am inclined to agree with Andrew, if for > no other reason than that those working on guest OSs do not generally > use debug hypervisors if they are not also working on Xen itself.
That's likely a mistake then, at least for work on PV guest OSes. In particular PV memory management code is sprinkled with gdprintk(), in an attempt to help those people understand why some operation has failed. Jan
