On 09/04/2019 10.35, Markus Armbruster wrote: > Thomas Huth <th...@redhat.com> writes: > >> On 09/04/2019 09.45, Markus Armbruster wrote: >>> Thomas Huth <th...@redhat.com> writes: >>> >>>> We'd like to get rid of global_qtest in the long run (since it is >>>> causing trouble for tests that run multiple instances of QEMU in >>>> parallel, e.g. migration tests)... so if it is feasible, please don't >>>> use it in new code anymore. Try to use a local variable in the function >>>> that call qtest_initf() and pass the test state around via a parameter >>>> to the functions that need it. >>> >>> Twenty tests still use @global_qtest >>> >>> Either we're serious about getting rid of @global_qtest. Then we should >>> just do it. >> >> Ha ha, "just do it" ... that's quite a bit of work, actually. It's not >> just about grep'ing for global_qtest, you also have to replace all the >> writel(), readl() etc. functions with qtest_writel(), qtest_readl() etc. > > And that's precisely why I'm reluctant to demand this work from > contributors. Asking nicely is of course fair.
That's what I did, didn't I? I said "... so if it is feasible, please don't use it in new code anymore". I did not say "you must not use this in new code anymore". So where's your problem here, Markus? Stephen, for the records, in case my mail was not clear: I'm fine if you keep the code as it currently is. But I'd be really happy if you could change it to avoid global_qtest. Thomas