On 13/09/20 17:51, Philippe Mathieu-Daudé wrote: >>> >> Yes, I'm not sure about the stub _in general_. I would prefer to either >> have the type available even on user-mode emulation, or remove the >> property altogether... I don't know, anything but not a property that >> aborts. :) > I plan to eventually get user-mode cleaned, but this is out of the scope > of this generic second part which is already big enough for Laurent to > digest. Maybe part #4 will be x86 specific. > > Let's wait for Markus view on this (personally I don't mind removing > the abort, it was not there on v3). >
The abort is better than v3 which was buggy (like use-uninitialized-data buggy). But I'd prefer not needing the stub at all. Paolo
