> On Oct 29, 2021, at 7:54 AM, Jason Thorpe <thor...@me.com> wrote: > > > >> On Oct 29, 2021, at 2:45 AM, Martin Husemann <mar...@duskware.de> wrote: >> >> On Thu, Oct 28, 2021 at 11:14:39AM -0700, Jason Thorpe wrote: >>> We really just need to kill the sigcontext stuff completely. More to the >>> point, we should version sigaction() before (or concurrently with) adding >>> whatever call we add here so as to ensure that it will only ever be a >>> ucontext. >> >> Something there is broken in -current, see PR 56471 (freshly compiled macppc >> ntpd tries to use compat_16___sigreturn14). > > Yes, and from your description it should have failed to register a > “sigcontext” handler in the first place.
Martin, is kern.module.autoload enabled on your machine? (Sorry, my Mac mini is in a state of “has been turned off for a while”, so I need to resurrect it today, so asking for info here to get the investigation going…) The PowerPC signal trampoline also has a case where it can return EINVAL because it thinks the saved SRR1 has been tampered with. I’m wondering if somehow the old sigcontext trampoline is getting invoked for a siginfo-style handler. -- thorpej