On Sun, May 3, 2009 at 8:56 PM, Marco S Hyman <marc@snafu.org> wrote:
quoted text
> On May 3, 2009, at 8:28 PM, Philip Guenther wrote:
>
>> recvfrom?  Are you no longer talking about
>> /usr/src/regress/lib/libpthread/dup2_race/dup2_race.c?
>
> /usr/src/regress/lib/libpthread/blocked_dup2/blocked_dup2.c

Hah, let me break out the sack cloth and ashes...


quoted text
> It blocks on call to recvfrom which is aborted when dup2 "closes"
> the second fd. I believe this test only has meaning in the pthread
> envoronment

I agree.  The behavior of this case was asked about on the
austin-group mailing list; the only response was this:

https://www.opengroup.org/sophocles/show_mail.tpl?CALLER=show_archive.tpl&sou
rce=L&listname=austin-group-l&id=5123

...which makes sense to me.  libpthread has the first behavior
mentioned there, while librthread has the third.


Philip Guenther