Re: kern/46464: lib/librumphijack/t_tcpip:ssh test case randomly fails

[Antti Kantee <pooka%NetBSD.org@localhost>]

The following reply was made to PR kern/46464; it has been noted by GNATS.

From: Antti Kantee <pooka%NetBSD.org@localhost>
To: Andreas Gustafsson <gson%gson.org@localhost>
Cc: gnats-bugs%NetBSD.org@localhost
Subject: Re: kern/46464: lib/librumphijack/t_tcpip:ssh test case randomly
 fails
Date: Sun, 20 Jul 2014 10:48:36 +0000

 On 20/07/14 08:41, Andreas Gustafsson wrote:
 > Antti Kantee wrote:
 >> I am almost sure you are correct.  The rest of the code seems to handle
 >> two polls finishing correctly, too.  That assert and the same one in
 >> another branch a few lines lower should be removed.
 >
 > I removed the asserts, but the test is still failing; now the "rv_host > 0"
 > assertion is triggering instead:
 >
 >    assertion "rv_host > 0" failed: file 
 > "/tmp/bracket/build/2014.07.20.00.48.51-i386/src/lib/librumphijack/hijack.c",
 >  line 2073, function "_sys___pollts50"
 
 Can NetBSD's poll() return with POLLIN set in revents, but not 
 reflecting that in the return value?  Is that legal behaviour?
 
 Or is there something else fishy going on?