Re: Missing in sigaction
Re: Missing in sigaction
- Subject: Re: Missing in sigaction
- From: Nat! <email@hidden>
- Date: Thu, 29 Mar 2012 13:45:15 +0200
Am 28.03.2012 um 09:07 schrieb Tilghman Lesher:
>
> You should be able to achieve the same effect (without the identified
> race condition) simply by adding SA_NODEFER to your sa_flags.
> _______________________________________________
OK, now this has been helpful and I think I now have a consistent theory about it:
If a signal arrives on a signal that is blocked, it's stored as pending. If another signal arrives while the other is still pending it gets dropped, because the kernel doesn't want to stack signals indefinetely. At this point though the contained information in struct __siginfo is almost useless, because - for instance - only one of the possibly different sending pids could be identified. Therefore it's actually better to not receive a struct __siginfo at all.
With SA_NODEFER, signals aren't blocked so the struct __siginfo can be passed unmodified.
If that were the case then the absence of struct __siginfo would actually be a valuable clue.
Ciao
Nat!
---------------------------------------------------
Java's the worst thing -- good way to make things a
tenth as fast as they should be. - J. Carmack
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Darwin-dev mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden