[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: SO_LINGER (ugh)
|
From: |
Roland McGrath |
|
Subject: |
Re: SO_LINGER (ugh) |
|
Date: |
Mon, 15 Oct 2001 00:56:20 -0400 (EDT) |
I agree with your analysis. I think your current behavior is probably best.
This is a general problem with all semantics that entail something active
or synchronized happening on `close', e.g. POSIX.1 file locking has similar
questions (though not quite as bad). As I read the Linux implementation, a
process dying (even by SIGKILL) will just stick around and block until the
(unbounded) linger timeout expires, before it reports death to its parent.
We will certainly never have behavior like that!