Weekly edition	Kernel	Security	Distributions	Contact Us	Search
Archives	Calendar	Subscribe	Write for LWN	LWN.net FAQ	Sponsors

Stupid question

Posted Feb 16, 2007 19:47 UTC (Fri) by spitzak (guest, #4593)
Parent article: Alternatives to fibrils

In both cases the system call is done by the parent process and the return is the child process. Why can't this be done the other way around, where the asyncrhonous call is being done by the child and it returns immediately to the parent? That would make a lot more sense, so I assumme there is a sensible reason that I can't figure out for how they are doing this.

(Log in to post comments)

Stupid question - that Linus anticipated

Posted Feb 17, 2007 2:05 UTC (Sat) by ds2horner (subscriber, #13438) [Link]

In his explanation it was so that the fork code could be reused with no modifications. He implies it could be done if it matters to callers.

Linus:
Now, I agree that this is a bit ugly in some of the details: in
particular, it means that if the system call blocks, we will literally
return as a *different* thread to user space. If you care, you shouldn't
use this interface, or come up with some way to make it work nicely (doing
it this way meant that I could just re-use all the clone/fork code as-is).

Stupid question - that Linus anticipated

Posted Feb 20, 2007 0:28 UTC (Tue) by mikov (subscriber, #33179) [Link]

How could it _not_ matter to callers ? If the thread id can change arbitrarily based on factors outside of the applications control - e.g. if some driver buffer is empty - then the thread id becomes completely pointless.

To me this seems completely unacceptable. Am I missing something ?

Stupid question - that Linus anticipated

Posted Feb 22, 2007 21:02 UTC (Thu) by huaz (guest, #10168) [Link]

You are right, it's indeed unacceptable.