Daniel P. Berrangé <berra...@redhat.com> writes: > On Tue, Oct 22, 2024 at 12:37:46PM +0200, Markus Armbruster wrote: >> Daniel P. Berrangé <berra...@redhat.com> writes: >> >> > On Tue, Oct 22, 2024 at 10:41:29AM +0200, Markus Armbruster wrote: >> >> Peter Xu <pet...@redhat.com> writes: >> >> >> >> > Per previous discussion [1,2], this patch deprecates >> >> > query-migrationthreads >> >> > command. >> >> > >> >> > To summarize, the major reason of the deprecation is due to no sensible >> >> > way >> >> > to consume the API properly: >> >> > >> >> > (1) The reported list of threads are incomplete (ignoring destination >> >> > threads and non-multifd threads). >> >> > >> >> > (2) For CPU pinning, there's no way to properly pin the threads with >> >> > the API if the threads will start running right away after >> >> > migration >> >> > threads can be queried, so the threads will always run on the >> >> > default >> >> > cores for a short window. >> >> > >> >> > (3) For VM debugging, one can use "-name $VM,debug-threads=on" >> >> > instead, >> >> > which will provide proper names for all migration threads. >> >> > >> >> > [1] https://lore.kernel.org/r/20240930195837.825728-1-pet...@redhat.com >> >> > [2] https://lore.kernel.org/r/20241011153417.516715-1-pet...@redhat.com >> >> > >> >> > Signed-off-by: Peter Xu <pet...@redhat.com> >> >> [...] >> >> >> > diff --git a/migration/threadinfo.c b/migration/threadinfo.c >> >> > index 262990dd75..2867413420 100644 >> >> > --- a/migration/threadinfo.c >> >> > +++ b/migration/threadinfo.c >> >> > @@ -13,6 +13,7 @@ >> >> > #include "qemu/osdep.h" >> >> > #include "qemu/queue.h" >> >> > #include "qemu/lockable.h" >> >> > +#include "qemu/error-report.h" >> >> > #include "threadinfo.h" >> >> > >> >> > QemuMutex migration_threads_lock; >> >> > @@ -52,6 +53,9 @@ MigrationThreadInfoList >> >> > *qmp_query_migrationthreads(Error **errp) >> >> > MigrationThread *thread = NULL; >> >> > >> >> > QEMU_LOCK_GUARD(&migration_threads_lock); >> >> > + >> >> > + warn_report("Command 'query-migrationthreads' is deprecated"); >> >> >> >> We don't normally do this for QMP commands. >> >> >> >> Management applications can use -compat deprecated-input=reject to check >> >> they're not sending deprecated commands or arguments. >> >> >> >> Suggest to drop. >> > >> > They could, but in practice I don't believe anything is doing this, so >> > the warning message is a practical way to alert people to the usage. >> >> Again, we not normally do this. What makes this one different? > > Do we not ? My expectation is that everything we record in deprecated.rst > also has a corresponding warn_report / warn_report_once in the code. > We know users may not read the docs, so we have a multi-pronged approach > to alerting them.
We definitely do not. >> Stepping onto my soapbox: if stuff going away surprisingly would cause >> you enough inconvenience to make early warning desirable, testing with >> suitable -compat is a lot more reliable than relying on warnings. >> *Especially* when your automated testing files warnings unexamined >> together with any other crap that may go to stderr, so your best chance >> to notice the warning is in ad hoc manual testing of QEMU. Nobody does >> that until after things broke. > > I don't see it as an either or choice. We try to surface the deprecation > info in as many different ways as is practical, as no single approach is > going to hit all bases. > > * Document it (deprecated.rst) > * Warn on QEMU stderr if used at runtime (warn_report) > * Enable apps to validate their usage in tests (-compat) > * Mark guests as tainted (libvirt API & VM log file, for certain asepts) To deprecate a QMP command, event, argument or result, we add the feature flag. One-liner, basically impossible to screw up. Review should then catch a missing update to deprecated.rst. Same for un-deprecating something. So far, this is as simple as it could possibly be. That's a feature. A warning requires additional handwritten code. Which *can* be screwed up. Moreover, missing warning (add or delete) is easy to miss in review. If we want such warnings for QMP, they should be automated just like the -compat actions. Any existing warnings rendered redundant should then be taken out. I considered that when I did -compat, and rejected it as not worth the effort.
