On 04/21/2013 01:12 PM, Lluís Vilanova wrote: > This interface provides two sets of operations: > > * Loading/unloading a trace instrumentation library. > > * Controls the instrumentation callbacks of the tracing events. > > Note that in the case of static instrumentation, the library is not > loaded/unloaded, but is still properly (de)initialized when QEMU starts and > exits.
I don't know if the series in general will be accepted, but if so, here's some interface things to think about: > +++ b/instrument/qapi-schema.json > @@ -0,0 +1,26 @@ > +# *-*- Mode: Python -*-* > +# > +# QAPI trace instrumentation control commands. > +# > +# Copyright (C) 2012-2013 Lluís Vilanova <vilan...@ac.upc.edu> > +# > +# This work is licensed under the terms of the GNU GPL, version 2 or later. > +# See the COPYING file in the top-level directory. > + > +## > +# @InstrType > +# > +# Instrumentation type supported by the system. > +# > +# @None: No instrumentation support. > +# > +# @Static: Static instrumentation support. > +# > +# @Dynamic: Dynamic instrumentation support. Generally, QMP enums are lower case (s/None/none/, and so forth). > +# > +# Warning: Keep in sync with #QIType. If we had the hypothetical tool that converted .json into end-user documentation, this sort of comment does not belong in the end-user document. Do we need to invent a syntax for development-only comments within .json files for things that are useful during qemu development? Or is it just better to move the comment to instead be in the C file that declares the enum, reminding any developer that modifies the enum to also modify the QMP type that reflects it? (We went with the latter approach in the recent query-command-line-options, putting the reminder comment to keep QMP and C in sync only in the C file.) > +# > +# Since: 1.5 This missed 1.5; it will need to be 1.6 now. -- Eric Blake eblake redhat com +1-919-301-3266 Libvirt virtualization library http://libvirt.org
signature.asc
Description: OpenPGP digital signature
