On Wed, Apr 08, 2015 at 10:30:52AM +0200, Paolo Bonzini wrote: > > > On 07/04/2015 22:46, Eduardo Habkost wrote: > > This uses the feature name arrays to register "feat-*" QOM properties > > for feature flags. This simply adds the properties so they can be > > configured using -global, but doesn't change x86_cpu_parse_featurestr() > > to use them yet. > > Out of curiosity, why the prefix? (Also, perhaps a prefix such as > "cpuid-*" would be better since the property often only affects the > cpuid leaves, rather than the availability of the feature itself).
The prefix exists to allow those properties to be easily identified by software that doesn't know the full list of feature names (I even took advantage of that on the x86-cpu-model-dump script). About the reason for using the "feat-" prefix, the short answer is "it's the prefix used in the last patch that implemented this (by Igor)". I think the first suggestion was to use "f-", then we changed to "feature-" or "feat-", and simply stayed using "feat-" in the last few versions. But I like the "cpuid-" suggestion and plan to use it on v2. Any objections? -- Eduardo
