This whole series looks fine to me. I think in commenting and cleaning up any of this, it bears explicit mention that (almost) every signal is potentially reduced to SI_USER. That is, in siqueue exhaustion you don't get any info and only non-special non-SI_USER >=SIGRTMIN signals ever fail to get posted, so you get the all-zeros defaults from collect_signal() at delivery time. That's the principle you're encoding in your si_fromuser() logic, but your logs and comments are not explicit about the relationship between that logic and what's implicit in the queue-exhaustion behavior. Thanks, Roland _______________________________________________ Containers mailing list Containers@xxxxxxxxxxxxxxxxxxxxxxxxxx https://lists.linux-foundation.org/mailman/listinfo/containers
