Most of these patches are a result of syz-kaller runs.

With these patches I get a significant reduction in fail, but we're not there
yet.  There is at least one known issue with unthrottling events (if only it
wouldn't not go into hiding whenever I add more debug code), and there are a
number of NMI watchdog triggers that so far make little sense.

There's a number of patches I'm not really confident in, so please have
a hard look at all of them.

Reply via email to